Buku JMTI_Kosen Hiroshima

2025 INTERNATIONAL SEMINAR Multidisciplinary Research and Distinct Culture: A Compilation of Final Year Projects from Japan-Malaysia Technical Institute (JMTI) & Hiroshima Kosen Japan-Malaysia Technical Institute Manpower Department Ministry of Human Resources

2025 INTERNATIONAL SEMINAR Multidisciplinary Research and Distinct Culture: A Compilation of Final Year Projects from Japan-Malaysia Technical Institute (JMTI) & Hiroshima Kosen First Edition 2026 Copyright © 2026 by Institut Teknikal Jepun-Malaysia All rights reserved. No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval system, without the prior written permission of the copyright owner and the publisher.eISBN 978-629-94724-0-7 Publisher: Institut Teknikal Jepun-Malaysia Manpower Department Ministry of Human Resources 59, Lorong Perindustrian Bukit Minyak 15, Taman Perindustrian Bukit Minyak, 14100 Simpang Ampat, Seberang Perai Tengah, Pulau Pinang Tel: 04-5087800 Fax: 04-5087809 Website: www.jmti.gov.my

i CONTENTS PREFACE…………………………………………………………………………………………… ii FOREWARD Ts. Hj. Zulkifle Bin Omar, Director, JMTI………………………………………………………….iii ORGANIZING COMMITTEE FOR 2025 INTERNATIONAL SEMINAR IN MULTIDISCIPLINARY RESEARCH AND DISTINCT CULTURE……………………………...iv TECHNICAL AND ENGINEERING RESEARCH CATEGORY………………………………….1 Development of a Mini CNC Machine for Educational Purposes at JMTI…………………………..2 Cylinder Head Innovative Fixture (CHEIF1824)…………………………………………………….7 Predictive Maintenance of a Centrifugal Pump Using Vibration Analysis………………………….11 Development of Waste Separation System with Microcontroller System Integration as Learning Material……………………………………………………………………………………15 Integrated Leaked Gas Detector with Voice Alert and IoT System…………………………………19 Remote Controlled Bionic Hand…………………………………………………………………….23 Graphical User Interface (GUI) - Based OCR Application for Text Recognition…………………..27 Comparative Design and Analysis of Micro-strip Patch Antenna Geometries for Non-Invasive Sweat Based Dielectric Sensing at 2.4 Ghz…………………………………………………………33 A Study on Combustion Characteristics of Water-Emulsified with New Generation Bio Diesel Fuel ………………………………………………………………………………………………….38 Analysis of Warm Up and Cold Down Processes in Marine Engine Operation Training…………..39 Honda's Engineering Marvels: The Pride of Japan………………………………………………….40 Graduation Research in Technical College Education using OpenFOAM………………………….41 CULTURAL STUDIES CATEGORY……………………………..……………………………….42 Similarities and Differences between Japanese and Malaysian Cultures…………………………….43 Japanese Etiquette and Manners……………………………………………………………………..44 The Origins, History and Present State of Kendo……………………………………………………45 A Glimpse into Malaysia’s History and Culture…………………………………………………….46 CREATIVE OR PERSONAL INTEREST PROJECTS (HOBBIES) CATEGORY ………………47 One-Month English Study in the Philippines………………………………………………………..48 The Appeal of Guitar ………………………………………………………………………………. 49

ii PREFACE 2025 International Seminar in Multidisciplinary Research and Distinct Culture The 2025 International Seminar on Multidisciplinary Research and Distinct Culture was held on 19th December 2025 at the Japan-Malaysia Technical Institute (JMTI). Organized by JMTI and attended by participants from both the host institution and Hiroshima Kosen, Japan, the program served as a dedicated platform for students to present the findings of their Final Year Projects (FYP) or Group Research Projects (GRP). The objectives of the seminar are as follows: i) Encourage research, development, and innovation among students through the presentation of studies, projects, and problem-solving approaches across various disciplines. ii) Support, promote, and enhance scholarly publications by technical institutions in multidisciplinary fields. iii) Facilitate the sharing of knowledge, skills, and expertise through seminar presentations among both local and international institutions. As the inaugural edition of this seminar, the 2025 program marks a significant milestone in establishing academic and skills cooperation between JMTI and Hiroshima Kosen. It provides a new avenue for research engagement among students at both institutions while laying the foundation for future international collaboration. A total of 18 papers were accepted for inclusion in this publication, with 9 papers from JMTI and 9 papers from Hiroshima Kosen. These papers cover areas that include, but are not limited to:  Technical and Engineering Research  Cultural Studies  Creative or Personal Interest Projects (Hobbies) The editors commend all contributors for their dedication and intellectual effort in producing high-quality papers for this publication. In addition, sincere appreciation is extended to all reviewers and organizing committee members from both institutions, whose unwavering support ensured the success of the 2025 International Seminar.

iii FOREWORD Ts. Hj. Zulkifle Bin OmarDirector Of Japan-Malaysia Technical Institute (JMTI)2025 International Seminar in Multidisciplinary Research and Distinct Culture Greetings and warm regards, The Japan-Malaysia Technical Institute (JMTI) is proud to have successfully organized the 2025 International Seminar on Multidisciplinary Research and Distinct Culture on 19thDecember 2025. The seminar was held at JMTI and attended by participants from both the host institution and Hiroshima Kosen, Japan. I would like to express my sincere gratitude to all parties involved in making this seminar a success. This seminar provided a valuable platform for students to present their Final Year Projects (FYP) or Group Research Projects (GRP), highlighting research, innovation, and problem-solving across diverse disciplines. A total of 18 papers were accepted for inclusion in this publication, representing the efforts of students from JMTI and Hiroshima Kosen. The seminar covered a wide range of areas, including technical and engineering research, cultural studies, and creative or personal interest projects The successful organization of this first edition marks an important milestone in fostering academic collaboration and skill development between JMTI and Hiroshima Kosen. It is our hope that future editions will attract participation from an even broader range of institutions, both locally and internationally, further enhancing research engagement and knowledge exchange in technology, engineering, and multidisciplinary studies. Finally, I wish to express my sincere appreciation to all contributors, reviewers, and organizing committee members for their dedication in ensuring the smooth conduct of this seminar and the publication of the project papers Thank you. (TS. HJ. ZULKIFLE BIN OMAR) DIRECTOR, JAPAN-MALAYSIA TECHNICAL INSTITUTE, MANPOWER DEPARTMENT, MINISTRY OF HUMAN RESOURCES

iv 1 ORGANIZING COMMITTEE FOR 2025 INTERNATIONAL SEMINAR IN MULTIDISCIPLINARY RESEARCH AND DISTINCT CULTURE CHAIRPERSON: Ts. Hj. Zulkiflie bin Omar VICE-CHAIRPERSON: En. Nuzul Ridzuan bin Padzil SECRETARY: Ts. Aini Salwa binti Hasan Nudin PROGRAM STEERING COMMITTEE: 1. Ts. Hj. Ahmad Muzafar bin Mamat @ Adam 2. Ts. Shaik Mohamed bin Mohamed 3. Ts. Dr. Noor Azam bin Ja’afar 4. Ts. Zuraini binti Ghazali AWARDS & CERTIFICATES COMMITTEE: 1. Ts. Syazwin binti Ahmad @ Ahmad Sowi 2. En. Muhamad Nazri bin Che Ani 3. Tc. Shalihin bin Abdullah BANQUET AND REFRESHMENT COMMITTEE: 1. En. Mohamad Najib bin Shaari 2. En. Ahmad Amru bin Mat Saad 3. Ts. Mohamad Danuri bin Suparman TECHNICAL AND VENUE PREPARATION COMMITTEE: 1. Tn. Hj. Yusri bin Md. Taib 2. Ts. Nor Yusuhanna binti Mohd Yusoff 3. En. Ayub bin Mohd Rasidi 4. En. Azman bin Supa 5. Tc. Fakharudin bin Mohd Yusof 6. Pn. Noor Hayanti binti Ahmad Hashim 7. Pn. Shakirah binti Yatim 8. Pn. Siti 'Aisyah binti Samat 9. Pn. Haslinda binti Nordin 10. Pn. Nor Fadzilah binti Abd Rahman PROTOCOL AND MEDIA COMMITTEE: 1. En. Ahmad Shafarin bin Shafie 2. Dr. Hoo Seng Chun 3. En. Kamal Aqili bin Shafie 4. Mr. Obi Kazuoki

v LOGISTICS AND ACCOMMODATION COMMITTEE: 1. Pn. Haffizawati binti Zini 2. Pn. Norshahira binti Din 3. En. Mahazan bin Baharom 4. En. Mat Shokri bin Zakaria EDITORIAL AND ADVERTISING COMMITTEE: 1. Ts. Dr. Noor Azam bin Ja’afar 2. Ts. Shaik Mohamed bin Mohamed Yusoff 3. Ts. Mohd Faisal bin Othman 4. En. Mohd Hezri bin Abdullah 5. En. Ahmad Shafarin bin Shafie 6. Ts. Mohamad Danuri bin Suparman 7. Ts. Dr. Amer Isyraqi bin Hussin 8. Pn. Wan Rohanina binti Wan Ibrahim COMPILATION AND BOOK PUBLICATION COMMITTEE: 1. Dr. Hoo Seng Chun 2. En. Ahmad Amru bin Mat Saad 3. Ts. Ong Joo Hun 4. En. Zambri bin Abdul Halim 5. En. Kamal Aqili bin Shafie 6. En. Muhammad Asnoor bin Abd Abas 7. Ts. Zuraini binti Ghazali

1 TECHNICAL AND ENGINEERING RESEARCH CATEGORY

2 Development of a Mini CNC Machine for Educational Purposes at JMTINoor Azam Jaafar1, *, Ahmad Amru Mat Saad1, Mazni Tajudin1, Risharwin Bala1 1Department of Mechanical & Production Engineering Technology, Manufacturing Engineering Technology Division, Japan-Malaysia Technical Institute, 59, Lorong Perindustrian Bukit Minyak, 15, Kawasan Perindustrian Bukit Minyak, 14100, Simpang Ampat, Penang Malaysia *Corresponding Author Email: [email protected] 1. Introduction The modern manufacturing sector places a high premium on proficiency in Computer Numerical Control (CNC) technology, making it a critical component of technical and vocational education training (TVET) curricula [1]. The Japan-Malaysia Technical Institute (JMTI) is a leading technical institution tasked with providing highly skilled manpower in manufacturing technology, emphasising the use of current technology in teaching. CNC machines are controlled automatically by a computer, allowing for the creation of products with specified shapes and sizes. The widely used modern methods of CAD/CAM and CNC are essential in the global manufacturing sector [2]. However, commercial CNC machines are typically large, complex, and expensive, which makes it difficult for educational and technical training institutions to acquire a sufficient number of units for hands-on learning. A major challenge in CNC education is the high cost of ownership and maintenance of full-sized industrial CNC machines, which also require large operating spaces and significant energy use [3]. Therefore, there is a clear necessity to develop a Mini CNC Machine that is more affordable yet still technically relevant for educational purposes. This compact machine must maintain the essential functions of CNC while offering a practical and cost-effective solution for technical education and training centres. The project's goal is to provide a compact and portable training tool to introduce students to the fundamental concepts of CNC machining. The specific objectives of this project are to design a Mini CNC Machine using the CAD software Autodesk Inventor, to fabricate the parts of the Mini CNC Machine using existing machines in the JMTI Manufacturing Workshop, and to analyse and validate the functionality and performance of the developed Mini CNC Machine. 2. Literature review 2.1 The evolution and role of CNC technology The history of Computer Numerical Control (CNC) machines traces back to the mid-20th century with the development of Numerical Control (NC) systems, a concept pioneered in the 1940s by John T. Parsons to improve the precision of complex component manufacturing, particularly for the aerospace industry. The success led to the first NC milling machine developed in 1952 by Richard Kegg in collaboration with the Massachusetts Institute of Technology (MIT), which used punch cards to control cutting tool movements and achieve unprecedented accuracy. The subsequent integration of microprocessors gave rise to modern CNC systems. Today, CNC technology is foundational to modern manufacturing, enabling high levels of efficiency, accuracy, Abstract: The continuous advancement of modern manufacturing necessitates highly skilled personnel proficient in Computer Numerical Control (CNC) technology, making it a crucial focus in technical and vocational education training (TVET). However, the adoption of hands-on training is often hampered by the significant capital investment, large footprint, and complex maintenance associated with commercial, industrial-grade CNC machines. This project addresses this critical gap by reporting the design, fabrication, and performance evaluation of a Mini CNC Machine intended as an affordable, compact, and effective training tool for the Japan-Malaysia Technical Institute (JMTI). This project demonstrates an integrated, low-cost CNC prototype utilizing open-source GRBL control optimized for compact, accessible educational applications. The methodology employed a systematic, phased approach, beginning with a comprehensive literature review on control systems and mechanical components. Mechanical and electrical designs were precisely engineered using Autodesk Inventor, followed by the fabrication of structural components from Aluminium 6061 using in-house resources such as CNC Milling and WEDM, with a final anodizing process for enhanced durability. The control system was built upon an Arduino board operating GRBL firmware, enabling G-code compatibility and simplified operation. The performance validation included three key assessments: motion accuracy, Manual Data Input (MDI) functionality, and cutting accuracy. The motion accuracy tests, measured using a 0.01 mm dial gauge, revealed an acceptable average motion error, with the maximum deviation observed at ± 0.05 mm on the Z-axis, primarily due to inherent ball screw backlash. Functionality tests confirmed the system’s immediate, error-free response to MDI commands. Crucially, the cutting accuracy test on acrylic material demonstrated excellent precision against the 15 mm × 20 mm target, yielding minimal average dimensional errors of ± 0.01 mm. These results validate the machine’s effectiveness for basic CNC programming, rapid prototyping, and educational light-duty machining, confirming its potential to significantly enhance practical training accessibility at JMTI.Keywords: Mini CNC Machine, Technical Education, Arduino GRBL Control System

3 and automation. It allows for the production of components with tight tolerances that are difficult to achieve manually, ensuring superior product consistency and quality. The process is driven by G-code, generated either through manual input or Computer-Aided Manufacturing (CAM) software. The widespread use of Computer-Aided Design (CAD) and CAM methodologies, coupled with CNC, is crucial in global manufacturing. In line with Industry 4.0, CNC technology continues to evolve, integrating features that enhance flexibility and intelligence. Recent advancements include the integration of the Internet of Things (IoT) for real-time machine condition monitoring and predictive maintenance, the use of Artificial Intelligence (AI) and Machine Learning algorithms to optimise cutting parameters and detect tool wear automatically, and the adoption of multi-axis processing (beyond 5 axes) for complex geometries [4]. These developments are crucial in sectors like automotive and aerospace, which demand maximum production efficiency and high precision. 2.2 The Role of Mini CNC Machines in Technical EducationModern technical education, particularly in engineering and manufacturing, necessitates hands-on experience with CNC technology. However, traditional industrial CNC machines present significant challenges for training institutions due to their large size, complexity, and high acquisition and maintenance costs. The high financial and operational burden often restricts the number of machines available for student use [5]. In response, the development of Mini CNC Machines provides a crucial and cost-effective alternative platform. These compact systems retain the fundamental operational principles of industrial machines but are significantly more affordable, space-saving, and mobile. This accessibility enables more institutions to integrate practical training, allowing students to gain experience in CNC programming and CAD/CAM concepts effectively. The machine also offers a safe and flexible training environment [6][7]. The feasibility of developing such small-scale machines has been bolstered by technological progress and the increased availability of affordable, standard components, such as stepper motors and ball screws. Furthermore, open-source control software like GRBL, running on platforms like Arduino, provides a robust and low-cost solution for managing G-code commands and controlling axis movement. While lower in cost than industrial PLC systems, the precision and capability of 3-axis Mini CNC machines using these controllers are deemed suitable for programming and operation training in technical institutions [8]. Applications of Mini CNCs extend beyond education to include small-scale prototyping and light machining in small industries (SMEs). 3. MethodologyThe project methodology applies a systematic and sequential approach, mirroring the Waterfall Design Model to ensure efficient execution and validation against the set objectives. This structured process was divided into several distinct phases, each completed and validated before proceeding to the next: (i) Planning and Requirement Analysis (incorporating the Literature Review), (ii) Design (Mechanical and Electrical), (iii) Development and Implementation (Fabrication and Assembly), and finally, (iv) Testing and Performance Evaluation (Analysis). As Figure 1 illustrates, this sequential structure enables each development phase to be implemented in an organized and focused manner. Fig. 1 - Waterfall Design Model 3.1 Design and fabrication The mechanical design phase was crucial, utilising Autodesk Inventor for 3D modelling, assembly drawings, and Bill of Materials (BOM) creation. Aluminium was selected for the frame components due to its strength and resistance to vibration. The motion system incorporated stepper motors, ball screws, and linear rails to ensure smooth, high-precision movement and minimal backlash. As a key part of the design documentation, Figure 2 outlines the Bill of Materials (BOM), providing a detailed inventory of all necessary materials and components for manufacturing the Mini CNC Machine. Fig. 2 - BOM for Mini CNC Machine The electrical design utilized an Arduino controller equipped with GRBL firmware due to its user-friendly control and compatibility with G-code [9][10]. The system incorporated A4988 motor drivers, an appropriate power supply to ensure stable operation, and safety components such as limit switches, as illustrated in Figure 3. The fabrication and material production phase involved machining the major components using JMTI's resources, including CNC Milling, CNC Turning, and WEDM. To ensure high-quality and longevity of the final product, a final anodizing process was performed on the aluminum parts to enhance surface durability and corrosion resistance

4 Fig. 3 - Electrical design circuit3.2 Testing and performance evaluation The assembled Mini CNC Machine underwent testing to assess its accuracy, repeatability, and cutting effectiveness. 3.2.1 Motion Accuracy Test The Motion Accuracy Test was essential to check how reliably the stepper motors and control system performed when commanded to move. The main goal was to quantify the amount of error and consistency (repeatability) as the machine moved between points. To conduct this test, the machine was instructed to move specific distances 5 mm, 10 mm, and 15 mm along the X, Y, and Z axes using Manual Data Input (MDI) commands. After each movement, the actual physical position of the axis was carefully measured and recorded using a precise dial gauge with a resolution of 0.01 mm. This direct measurement method allowed us to calculate the movement error and confirm the machine’s foundational precision before any cutting operations began. This method was previously applied in the development of a mini CNC machine as reported by Salam et al. [11]. Figure 4 illustrates the setup, showing the position of the dial gauge during the accuracy measurement of the X, Y, and Z axes. Fig. 4 - Position of the dial gauge during the movement accuracy test for the X, Y, and Z axes. 3.2.2 Manual Data Input (MDI) Test The Manual Data Input (MDI) Test was conducted to thoroughly check the control system's stability and responsiveness, particularly before running automated G-code programs. The test aimed to ensure that the machine could be reliably controlled directly by the operator. The procedure involved manually entering key commands, such as movement instructions (G00 for rapid travel, G01 for linear feed) and spindle controls (M03 to start, M05 to stop, S500 for speed setting). Observation focused specifically on the machine's immediate reaction time to these inputs. This test was crucial for verifying that the machine's control components, including the Arduino and GRBL firmware, responded accurately and without any delay or error when given simple, direct instructions. 3.2.3 Cutting Accuracy Test The final assessment involved the Cutting Accuracy Test, which was performed to confirm the machine's ability to cut precisely according to the engineering drawings. As shown in Figure 5, a rectangular shape measuring 15 mm x 20 mm was milled onto an acrylic sample. Machining parameters included a feed rate of 400 mm per minute, a cutting depth of 0.5 mm per pass, and a spindle speed of 2,000 revolutions per minute (RPM). After cutting, the actual dimensions of the sample were measured using the open-source software ImageJ. This digital method was chosen over tools like calipers because it provides more consistent and higher-precision results. The actual cutting error was then calculated by comparing the dimensions measured by ImageJ against the original drawing specifications. Fig. 5 - Engineering drawing for cutting accuracy test (15 mm x 20 mm rectangle).4. Result and discussion4.1 Motion Accuracy Test Results The motion accuracy test was performed to assess the machine's capability to achieve desired coordinates reliably. The test demonstrated that the movement error did not exceed 0.05 mm, which is considered within the acceptable range for educational and light machining applications. The average errors recorded were ±0.03 mm (X-axis), ±0.04 mm (Y-axis), and ±0.05 mm (Z-axis). Fig. 1 presents the detailed data from the test, showing the difference between the commanded movement distance and the actual measured reading for each axis. Figure 6 provides a visual representation of this data, illustrating the movement error versus distance for the X, Y, and Z axes. The errors were primarily related to technical factors such as backlash in the ball screw and precision limitations in the linear rail assembly.

5 Table 1 - Results of the motion accuracy test Fig. 6 - Movement Error versus Distance4.2 Manual Data Input (MDI) Test ResultsThe MDI test was conducted to verify the stability and responsiveness of the control system before executing automated programs. This capability is critical, as it allows the operator to effectively test axis movements and spindle functionality directly, minimizing the potential for errors. Table 2 summarises the outcome of the MDI commands. The test confirmed that the machine's control system functioned well, successfully responding to all axis movements and spindle controls without delay or error. Specifically, movement along the X, Y, and Z axes was successful with no delay, and the spindle control was successfully started and stopped. The overall result indicated no movement errors, validating the robustness of the Arduino/GRBL control interface. Table 2 - Manual Data Input (MDI) Test ResultsFunction Result Axis movement (X, Y, Z) Successful with no delaySpindle control Successfully started and stoppedMovement errors None4.3 Cutting Accuracy Test Results The cutting test involved milling the designated 15 mm x 20 mm rectangular geometry onto an acrylic sample using the defined machining parameters (Feed rate: 400 mm/min, Spindle speed: 2,000 RPM). The actual dimensions of the five cut samples were measured digitally using ImageJ software. Table 3 presents the detailed measurement data, and Figure 7 provides a visual comparison of the actual cut sizes against the ideal dimensions. The average actual dimensions recorded were 14.99 mm for the X-axis and 20.01 mm for the Y-axis. This resulted in a minimal average cutting error of 0.01 mm (X-axis) and 0.01 mm (Y-axis), confirming a minimal deviation from the original specification. Table 3 - Results of the motion accuracy test Sample Drawing Dimension (mm)Actual Meas. (X) (mm)Actual Meas. (Y)(mm)Error X (mm)Error Y (mm)1 15 x 20 14.96 20.02 0.04 -0.02 2 15 x 20 15.01 19.98 -0.01 0.02 3 15 x 20 14.97 19.99 0.03 0.01 4 15 x 20 15.02 20.02 -0.02 -0.02 5 15 x 20 14.98 20.05 0.02 -0.05 Ave. Errors (mm)- 14.99 20.01 0.01 -0.01 Fig. 7 - Movement Error versus Distance5. Conclusion and recommendation5.1 Conclusion The project resulted in the successful development of the Mini CNC Machine, meeting its primary objectives. Performance testing demonstrated that the machine operates with precision suitable for light machining applications. The following summarizes key test findings: Motion Accuracy: Testing indicated that movement error remained within ±0.05 mm, which is considered acceptable. Identified sources of error included backlash in the ball screw and imprecision in linear rail assembly. Manual Data Input (MDI) Performance: The control system showed a good response to manual input without any delays or errors in movement. This allows the operator to effectively test and control the machine before the actual cutting process. Cutting Accuracy: Cutting tests on acrylic material demonstrated the machine's ability to produce geometric shapes with minimal error. The achieved accuracy is suitable for prototype machining. In conclusion, the Mini CNC Machine fulfills the study’s objectives and is appropriate for basic CNC education and training. 5.2 Implications and Recommendations This study provides important practical implications and suggests a path for future improvements: i. Implications for Education and Industry: The machine is a cost-effective learning tool for technical training, allowing more students to gain practical CNC experience. It is also useful for Small and Medium-sized Enterprises (SMEs) requiring low-cost light machining. 0.020.05-0.02 -0.02-0.030.04 0.030.08-0.01-0.04-0.0200.020.040.060.080.1 Ave errors (mm) 0 5 10 15 20Distance (mm)Movement Error versus DistancePaksi XPaksi YPaksi Z14.96 15.01 14.97 15.02 14.9820.02 19.98 19.99 20.02 20.0514151617181920210 1 2 3 4 5 6 Actual Measurement (mm)Sample 1520Distance (mm) X axis (mm)Y axis (mm)Z axis (mm)5 4.98 5.02 4.97 10 9.95 10.03 9.92 15 15.02 14.96 15.01 Ave. error (mm) ±0.03 ±0.04 ±0.05

6 ii. Recommendations for Enhancement: a) Improving Movement Accuracy: It is recommended to use higher-grade ball screws to reduce backlash and to install a position encoding system (encoder) to improve the machine's ability to read axis movement. b) Enhancing Machine Structure: To reduce vibration during machining, the machine frame material should be improved, and the installation of the linear rails needs to be more precise. iii. Future Development: This study should be used to support further research into improving machine performance, including developing advanced control systems and integrating with Industry 4.0 technology. 5.3 Summary The research has demonstrated that the Mini CNC Machine has great potential for light machining and technical education. The accuracy levels achieved show that the machine can be used for training and small-scale production. However, there is still room for improvement in accuracy and mechanical structure. Further research is needed to strengthen this technology so it can compete with more advanced machining equipment. Acknowledgement The authors would like to express their sincere appreciation to JMTI, for the financial support provided through the Dana Inovasi JMTI, which made the development and testing of the mini CNC machine possible. References [1] T. T. Tung, N. X. Quynh, and T. V. Minh, “Development and Implementation of a Mini CNC Milling Machine,” Acta Marisiensis. Ser. Technol., vol. 18, no. 2, pp. 24–28, 2021, doi: 10.2478/amset2021-0014. [2] D. Santosh Thakur, S. Mohammed Shafi Shaikh, F. Salim Shaikh, B. Ganesh Pangam, M. Karim Khan, and M. R. Desai, “Design and Fabrication of LowCost Mini Cnc Milling Machine,” Int. Res. J. Mod. Eng. Technol. Sci., no. 04, pp. 2582–5208, 2023, [Online]. Available: https://www.doi.org/10.56726/IRJMETS37678 [3] S. M. Ali and H. Mohsin, “Design and Fabrication of 3-Axes Mini CNC Milling Machine,” IOP Conf. Ser. Mater. Sci. Eng., vol. 1094, no. 1, p. 012005, 2021, doi: 10.1088/1757-899x/1094/1/012005. [4] T. Katduang, D. Maneetham, P. N. Crisnapati, and W. Srichaipanya, “Integrating Low-Cost Mini CNC Machines with IoTEnabled Energy Monitoring and Machine Learning for Sustainable Manufacturing,” Int. J. Eng. Trends Technol., vol. 72, no. 6, pp. 82–91, 2024, doi: 10.14445/22315381/IJETTV72I6P109. [5] N. A. Ismail et al., “Mini CNC Training Kit,” Ski. Malaysia J., vol. 7, no. 1, pp. 5–8, 2022. [6] G. E. Pramono, “Rancang Bangun CNC Mini Router 3 Axis untuk Keperluan Praktikum CAD/CAM,” Tugas Akhir Progr. Stud. Tek. Mesin, 2015. [7] E. a. Amala, “Pengembangan Perangkat Lunak Sistem Operasi Mesin Milling CNC Trainer,” J. Tek. Mesin, vol. Vol.2, no. No.3, 2014. [8] R. Ahammed and T. Islam, Design , Contraction & Perform Test of Three Axis CNC Milling Machine Design , Contraction & Perform Test of Three Axis CNC Milling Machine, no. January. 2024. [9] B. B. Barik, A. Mahanty, S. D. Majumder, and A. Roy Goswami, “Fabrication of Cost-effective Threeaxis portable mini-CNC milling Machine,” Mater. Today Proc., no. xxxx, 2023, doi: 10.1016/j.matpr.2023.03.012. [10] A. R. Sonawane, A. B. Rane, and D. S. S. Sudhakar, “Development of a3-Axis Cnc Milling Machine With an Open Source Controller,” Int. J. Res. Eng. Technol., vol. 06, no. 08, pp. 9–15, 2017, doi: 10.15623/ijret.2017.0608002. [11] A. Salam, M. Iswar, M. Rifaldi, S. Malik, and K. Putra, “Rancang Bangun Mesin CNC Router Mini Untuk Pembelajaran Mahasiswa Jurusan Teknik Mesin,” J. Tek. Mesin Sinergi, vol. 17, no. 2, p. 150, 2020, doi: 10.31963/sinergi.v17i2.2077.

7 Cylinder Head Innovative Fixture (CHEIF1824)Mohd Faisal Bin Othman1, *, Mohd Zaidi Bin Kasim1, Rosmawani Binti Ghazali1, Isnorzaidi Bin Ismail1 1Department of Mechanical & Production Engineering Technology, Manufacturing Engineering Technology Division, Japan-Malaysia Technical Institute, 59, Lorong Perindustrian Bukit Minyak, 15, Kawasan Perindustrian Bukit Minyak, 14100, Simpang Ampat, Penang Malaysia *Corresponding Author Email: [email protected]. Introduction The cylinder head is a critical and complex component of an internal combustion engine, as it directly affects engine performance and efficiency. Many workshops face difficulties in handling cylinder heads during maintenance tasks such as porting, valve grinding, and polishing. In small and mediumsized workshops, the lack of a dedicated fixture often leads to inefficiency, longer working times, and higher safety risks. To overcome these challenges, CHEIF1824 was developed as a stable, adjustable, and durable fixture specifically designed for Toyota engines with capacities between 1800cc and 2400cc.The successful development of this fixture is expected to result in measurable improvements in process time, enhanced safety, and greater consistency in machining operations, thereby addressing the prevalent efficiency gaps in SME automotive workshops 2. Methodology The development of CHEIF1824 was carried out in several stages. First, a requirement analysis was conducted at QASMI Auto Specialist to identify inefficiencies and safety issues faced during cylinder head maintenance. Next, the fixture was designed and modelled using CAD software to ensure ergonomic use with adjustable angles of 45° and 90°. Aluminum and mild steel were selected as the primary materials to provide strength and durability, and stability. A prototype was then fabricated using machining processes such as turning, milling, and wire cutting. Finally, the fixture was tested in workshop conditions, and user feedback was collected to guide iterative improvements in both design and functionality. Table 1 - Process and Machine No. Process Type of Machine 1 Cutting the workpiece Horizontal Band Saw 2 Turning cylindrical parts of workpiece Lathe Machine 3 Facing the workpiece Conventional Milling Machine 4 Making slots, counters, etc. CNC Milling 5 Drilling Upright Drilling & Tapping Machine Abstract: The Cylinder Head Innovative Fixture (CHEIF1824) is a purpose-built holder for Toyota cylinder heads (1800–2400 cc) that provides rigid, repeatable clamping during porting and valve-grinding operations. Fabricated from aluminum with mild-steel load points, the fixture delivers higher strength, better durability, and safer handling than placing heads directly on a bench. A universal multi-hole mounting pad allows flexible bolt-up to varied bolt patterns, reducing re-clamp time and improving datum consistency across stages. In workshop trials at QASMI Auto Specialist, CHEIF1824 reduced processing time by about 50% (from ~3 h to ~1.5 h per head) and doubled daily throughput from 3 to 6 units without additional labor, while maintaining consistent finish quality. These results indicate a practical, efficient, and cost-effective upgrade for small and medium workshops, improving ergonomics and safety as well as productivity.Keywords: Cylinder Head, Fixture Design, Automotive Innovation, Valve Grinding

8 Fig. 1 - (a). List of Components & Assembly3. Results and Discussion The implementation of CHEIF1824 significantly improved workshop efficiency. Testing at QASMI Auto Specialist demonstrated that the fixture reduced porting and valve grinding time by half, from about three hours to only 1.5 hours. This improvement not only shortened turnaround time but also enabled mechanics to handle more jobs in each day. As a result, daily output doubled from three to six units, clearly showing the fixture’s potential to enhance productivity in small and medium-sized workshops. In addition to efficiency, the fixture enhanced safety and work quality. By securely holding the cylinder head, CHEIF1824 eliminated the risk of slippage commonly encountered when the component is placed directly on a table. The adjustable angles of 45° and 90° provided more comfortable working positions, reducing fatigue and physical strain for mechanics. These ergonomic benefits allowed for greater precision and consistency in machining, ensuring higher quality results in porting and valve grinding operations. Fig. 2 - (b). Adjustable working angles for valve grinding, polishing, and similar processes The practicality of the fixture was further confirmed through user feedback. Mechanics reported that CHEIF1824 was easy to operate and did not require specialized training, making it suitable for immediate adoption in automotive workshops. Despite being made from durable materials such as aluminium and mild steel, the fixture remained affordable, with an estimated cost of RM700–RM1200. Calculations showed that the investment could be recovered within approximately ten days of use, making it a cost-effective solution for workshop owners. Fig. 3 - (c). Toyota Cylinder Heads (1.8L–2.4L) Comparison 3.1 Performance, Efficiency, and Compatibility The CHIEF1824 fixture maintained secure, slip-free clamping across repeated cycles (strength), showed no measurable wear on contact faces or threads (durability), reduced movement risk during operations (safety), enabled one-person, repeatable setups on common benches and tools (usability), and required only routine cleaning with standard fasteners (maintainability). Compared with conventional practice, operation times were roughly halved: valve grinding from 1.5–2.5 h to 0.75–1.25 h, valve/chamber polishing from 1–1.5 h to 0.5–0.75 h, and intake/exhaust porting from 2.5–4 h to 1.25–2 h, reducing total time per head from 5–8 h to 2.5–4 h (≈40–60% savings, depending on head condition and finish). Trials on Toyota AZ-family heads 1AZ (1.8 L), 1ZZ (2.0 L), and 2AZ (2.4 L) and derivatives confirmed correct seating, tool access, and full workflows without modification across 35 model applications spanning Toyota, Scion, Lexus, and Pontiac.

9 Table 2 - Performance criteria and evaluation of the CHIEF1824 fixture Criterion Evaluation Strength High – Withstands clamping forces and engine weight Durability Long-lasting under repeated use and varying workshop condition Safety Improved – Secure holding reduces risk of accidents Usability High – Easy to operate, adjustable and compatible Maintainability Minimal – Simple design with low maintenance requirements Table 3 - Process time comparison: conventional method vs. CHIEF1824 (hours)Process Conventional Time (h) Time with CHIEF 1824 (h) Time Saved Valve Grinding 1.5 – 2.5 0.75 – 1.25 50% Polishing (Valve & Chamber) 1 – 1.5 0.5 – 0.75 50% Porting (Intake & Exhaust) 2.5 – 4 1.25 - 2 50% Table 4 - Total estimated process time per cylinder head by method (hours)Method Total Time (h) Conventional 5 – 8 CHIEF1824 2.5 – 4 Table 5 - Engine TestedNo. 1AZ – 1800CC1ZZ – 2000CC2AZ – 2400CC1 Toyota Camry (2001 – 2006)Toyota Avensis (2000 – 2009)Toyota Camry (2002 – 2011) 2 Toyota RAV4 (2000 – 2010)Toyota Celica (1999 – 2006)Toyota RAV4 (2004 – 2008) 3 Toyota Ipsum (2001 – 2009)Toyota Corolla Verso (2001 – 2009)Toyota Highlander/Kluger (2001 – 2007)4 Toyota Avensis (2001 – 2009)Toyota MR2 Spyder (2000 – 2006)Toyota Harrier (2000 – 2008) 5 Toyota Avensis Verso (2001 – 2009)Toyota RAV4 (2000 – 2006)Toyota Ipsum (2001 – 2009) 6 Toyota Caldina (2002 – 2007)Toyota Wish (2003 – 2009) Toyota Alphard (2002 – 2015) 7 Toyota Gaia (2001 – 2004) Toyota Estima/Previa/Tarago (2000 – 2019)8 Toyota Opa (2000 – 2005) Toyota Matrix (2009 – 2014) 9 Toyota Allion (2001 – 2007) Toyota Corolla XRS (2009 – 2010) 10 Toyota Premio (2001 – 2007) Toyota Mark X Zio (2007 – 2013) 11 Toyota Noah/Voxy (2001 – 2007) Toyota Avensis (2000 – 2009) 12 Toyota Wish (2003 – 2009) Scion tC (2005 – 2010) 13 Toyota Nadia (1998 – 2000) Scion xB (2008 – 2015) 14 Pontiac Vibe (2009 – 2010)15 Lexus HS 250h (2010 -2014)16 Toyota Sai (2009 - 2017)TOTAL 13 6 16 Finally, the innovation and performance of CHEIF1824 received formal recognition. The fixture was awarded a Gold Medal at the CITEC 2025 competition and secured First Place in the JMTI Final Year Innovation Competition (Mechanical category). Its validation through competitive events and practical application at QASMI Auto Specialist underscores its value as a reliable, efficient, and innovative tool for the automotive industry. 4. Practical Innovation We added a multi-hole universal mounting pad to the CHIEF1824 fixture. The regular hole matrix (mix of through, threaded, and locator-pin bores) enables adapter free clamping for diverse cylinder-head bolt patterns, improving compatibility and datum repeatability. In AZ-family trials, this cut setup/re-clamp time by ~10–15 minutes per head, contributing to the overall 40–60% process-time reduction.

10 Fig. 4 - (d). CHIEF1824 Universal Cylinder-Head Fixture 5. Conclusion CHEIF1824 delivers a practical and measurable improvement to cylinder head maintenance in SMEs: it halves porting/valve-grinding time (≈3 h → 1.5 h) and doubles daily throughput (3 → 6 units) while improving safety, ergonomics, and machining consistency. Its stable, adjustable design (preset 45°/90°) and robust materials (aluminum and mild steel) make it easy to adapt without specialized training, and the projected payback period is short (≈10 days). Field use at QASMI Auto Specialist and independent recognition (CITEC Gold; JMTI FYP winner) further validates its utility. Overall, CHEIF1824 is a cost-effective, workshop-ready fixture tailored to Toyota 1.8–2.4 L engines that raises productivity and quality concurrently. Fig. 5 - (e). Workshop Use of CHEIF1824 Fixture References [1] Vizard, D. (2012). How to Port and Flow Test Cylinder Heads. North Branch, MN: CarTech. [2] Henrikson, E. K. (1973). Jig and Fixture Design Manual. New York: Industrial Press. [3] Muhammad Asyraf, M. A. G., Surniza, M. H., Muhammad Hazim, I., & Muhammad Luqman Hafizi, S. (2022). Valve Seating Grinder. Final Year Digest, Politeknik Sultan Mizan Zainal Abidin (PSMZA), Terengganu, Malaysia. [4] BookAuthority. (n.d.). Toyota Production System - selected resources. BookAuthority. [5] Google Books. (n.d.). Jig and Fixture Design. Google Books. [6] Qasmi Auto Specialist. (2025). Project introduction of CHEIF1824 [Video]. Facebook. [7] Qasmi Auto Specialist. (2025). Customer review of CHEIF1824 [Video]. Facebook. [8] Qasmi Auto Specialist. (2025). Rigidity test of CHEIF1824 [Video]. Facebook.

11 Predictive Maintenance of a Centrifugal Pump Using Vibration Analysis Zambri bin Abdul Halim1, *, Mohamad Fauzi Bin Abu Bakar 1, Fakhrul Azman Bin Mohamed 1, Azizi Bin Mat Shariff11Department of Mechanical & Production Engineering Technology, Mechatronic Engineering Technology Division,Japan-Malaysia Technical Institute, 59, Lorong Perindustrian Bukit Minyak, 15, Kawasan Perindustrian Bukit Minyak, 14100, Simpang Ampat, Penang Malaysia*Corresponding Author Email: [email protected]. Introduction Centrifugal pumps are essential components in industrial operations such as power generation, petrochemical processing, and water treatment systems, where their operational reliability directly influences plant availability and production continuity [6]. Unexpected pump failures can cause significant downtime, energy loss, and maintenance costs, with bearing degradation being one of the most common sources of mechanical breakdown in rotating machinery [5]. Conventional maintenance approaches, such as preventive or corrective maintenance, often fail to detect early signs of bearing deterioration before severe damage occurs. Therefore, predictive maintenance techniques, particularly vibrationbased condition monitoring, have become a preferred method for assessing equipment health and preventing unplanned failures [3]. By analyzing vibration signatures in the frequency domain, it is possible to identify characteristic defect frequencies that indicate specific fault types within bearings [4]. This study focuses on the vibration condition of Pump Bearing #4 (Free End, type 6321) in a 450-kW centrifugal pump system, emphasizing frequency-domain analysis and comparison with other bearings in the drive train. Special attention is given to identifying characteristic defect frequencies including Ball Pass Frequency Inner (BPFI), Ball Pass Frequency Outer (BPFO), Ball Spin Frequency (BSF), and Fundamental Train Frequency (FTF) to evaluate the severity and progression of bearing wear. 2. Literature ReviewThe characteristic fault frequencies of a rolling element bearing depend on its geometry, specifically the number of rolling elements, pitch diameter, rolling element diameter, and contact angle, as well as the shaft rotational speed. According to Randall and Antoni [4], the most important frequencies are defined as follows: Ball Pass Frequency of the Inner Race: BPFI (fi) = 2ZS [ 1 + PdBdcos θ ] (1) Ball Pass Frequency of the Outer Race: BPFO (fo) = 2ZS [ 1 - PdBdcos θ ] (2) Ball Spin Frequency: BSF (fr) = BdPd2S [ 1 - (PdBd) 2 (cos θ)2 ] (3) where, S = Revolution per second Z = Number of rolling elements Bd = Roller diameter in mm Pd = Bearing pitch diameter θ = Contact angle These frequencies provide a diagnostic signature for detecting and differentiating bearing defects. For instance, outer race defects typically generate strong vibration components at BPFO and its harmonics, whereas inner race defects are Abstract: Centrifugal pumps are critical components in industrial operations where unexpected bearing failures can lead to costly downtime and reduced system reliability. This study aims to diagnose the vibration condition of a 450 KW centrifugal pump system using predictive maintenance techniques. Vibration data were collected from Pump Bearing #4 (Free End, 6321) under steady operating conditions and analyzed using Time Waveform (TWF), Fast Fourier Transform (FFT) spectra, and calculated bearing fault frequencies—ball pass frequency outer race (BPFO), ball pass frequency inner race (BPFI), ball spin frequency (BSF), and fundamental train frequency (FTF). The results show repetitive impact patterns and spectral sidebands corresponding to the inner race fault frequencies, particularly between the 21× and 25× order range, indicating progressive defect development. The findings confirm advanced inner race degradation, where vibration analysis effectively detected fault progression before catastrophic failure occurred. This demonstrates that conditionbased vibration monitoring is a reliable diagnostic approach for identifying early-stage bearing damage in high-power rotating machinery. It allows maintenance teams to take timely corrective action, minimizing unscheduled downtime and optimizing plant reliability. Corrective bearing replacement and continuous online vibration monitoring are recommended to enhance system performance and support long-term predictive maintenance strategies for critical centrifugal pump systems. Keywords: Time Waveform (TWF), Fast Fourier Transform (FFT) spectra, ball pass frequency outer race (BPFO), ball pass frequency inner race (BPFI), ball spin frequency (BSF), and fundamental train frequency (FTF).

12 dominated by BPFI and sideband patterns around the rotational frequency [4]. Rolling element and cage defects exhibit frequencies corresponding to BSF and FTF, respectively. 3. Measurement Setup and Methodology The vibration measurement setup for the centrifugal pump system is summarized in Table 1. The test was conducted on a 450kW centrifugal pump operating at 989 RPM (≈ 16.48 Hz) with a closed six-blade impeller and a direct motor–pump coupling. The bearing under investigation was the pump free-end bearing (Location #4), identified as a 6321 deep-groove ball bearing that supports both radial and axial loads. A tri-axial accelerometer [2] was mounted vertically on the pump housing to measure vibration responses transmitted through the bearing structure. All measurements were taken under steady-state operating conditions to ensure reliable spectral data for subsequent fault-frequency identification and analysis. Table 1 - Measurement setup for centrifugal pump system Motor power 450 KWOperating Speed 989 RPM (~16.48 Hz)Pump Type Centrifugal (closed impeller, six blades)Coupling Motor–pump direct couplingBearing Inspected 6321 deep groove ball bearingLocation pump free end (Location #4)Vibration measurements were collected in axial, tangential, and radial directions using the Fluke 810 diagnostic analyzer. The calculated fault frequencies for the 6321 bearing are summarized in Table 1, with order values of FTF = 0.38, BPFO = 3.08, BPFI = 4.92, and BSF = 2.05. These values are expressed in multiples of the shaft rotational frequency (order), which allows direct comparison with vibration spectra (Tandon & Choudhury, 1999). Table 2 - Bearing 6321 fault frequencies Fault Frequency Formula Description Oder Value FTF (Fundamental Train Frequency) FTF = (1/2) × fr × ( 1 - (d/Dp) × cosθ ) Cage frequency; useful for detecting cage faults.0.38 BPFO (Ball Pass Frequency Outer Race)BPFO = (Z/2) × fr × ( 1 - (d/Dp) × cosθ ) Outer race defect frequency. 3.08 BPFI (Ball Pass Frequency Inner Race)BPFI = (Z/2) × fr × ( 1 + (d/Dp) × cosθ ) Inner race defect frequency. 4.92 BSF (Ball Spin Frequency) BSF = (Dp / 2d) × fr × [ 1 - ((d/Dp) × cosθ )² ]Ball spin frequency; indicates ball defects.2.05 Table 2 summarizes the key fault frequency formulas used in this study. The Fundamental Train Frequency (FTF) represents the cage rotational frequency and is primarily used to detect cage-related faults. The Ball Pass Frequency Outer Race (BPFO) and Ball Pass Frequency Inner Race (BPFI) correspond to defects occurring on the outer and inner raceways, respectively, while the Ball Spin Frequency (BSF) represents the rotation of the individual rolling elements, which is associated with ball surface damage. The calculated theoretical order values for the 6321 bearing were 0.38× for FTF, 3.08× for BPFO, 4.92× for BPFI, and 2.05× for BSF, where “×” indicates multiples of the shaft rotational speed. These order values serve as reference points in the vibration spectrum for identifying characteristic defect peaks. Fig. 1. Tri-axial acceleration sensor mounted on bearing #4 in vertical direction Figure 1 shows the vibration measurement setup on the centrifugal pump assembly. A tri-axial accelerometer was mounted on the vertical position of the free-end bearing housing (Pump Bearing #4), which supports the pump shaft and accommodates axial and thermal expansion during operation. . 4. Results and Discussion4.1 Time Waveform (TWF) Figure 2 shows the time waveform (TWF) signals for both axial and radial directions, indicating periodic impact patterns with a fundamental period of approximately 0.062 s, corresponding to the shaft rotational speed of around 16.48 Hz. The axial signal exhibits higher impulsive peaks compared to the radial direction, suggesting stronger thrust or axial loading transmitted through the bearing structure. The presence of repeated impacts in both axes implies localized bearing excitation, possibly due to inner-race or rollingelement surface irregularities that generate vibration responses synchronized with shaft rotation. Fig. 2 - Time waveform (TWF) signals for both axial and radial directions

13 4.2 High range spectrum analysis Figure 3 explains that the spectrum shows a dominant peak at 25× order with amplitude ~10.15 mm/s, matching harmonics of BPFI (Ball Pass Frequency Inner race). This indicates advanced inner race fault. The axial direction is most severe because thrust load from the impeller is transmitted through the inner race. Fig. 3 - Spectrum Axial (High Range) – Pump Bearing #4 (Free End, 6321) Figure 4 explains that the spectrum reveals a strong peak at 21× order with an amplitude of ~9.17 mm/s. Tangential vibration confirms bearing fault modulation due to spalling. Sidebands around 8.84×–9.21× orders indicate repeated impacts each time rolling elements pass the defect. High tangential amplitude shows the defect is in an advanced stage. Fig. 4 - Spectrum Tangential (High Range) – Pump Bearing #4 (Free End, 6321) Figure 5 explains the radial spectrum shows lower amplitude (~1.54 mm/s at 25× order) but exhibits broadband excitation up to 34× order. This broadband is typical of impact energy exciting structural resonances. Although less severe than axial/tangential, it confirms that impulsive energy is spreading across the bearing housing, indicating fault progression. Fig. 5 - Spectrum Radial (High Range) – Pump Bearing #4 (Free End, 6321) 4.3 Low-range spectrum analysis Figure 6 shown peaks detected at 3.05× (0.86 mm/s) and 6.1× (0.83 mm/s) orders fall within the Ball Spin Frequency (BSF) sideband region, indicating early impact modulation. Although amplitudes are smaller compared to high range, the sideband pattern is an early symptom of inner race defect development. This shows fault initiation before high-order harmonics dominate. Fig. 6 - Spectrum Axial (Low Range) – Pump Bearing #4 (Free End, 6321) Figure 7 shows peaks at 8.84× (1.60 mm/s) and 9.21× (1.45 mm/s) orders align with BPFI (Ball Pass Frequency Inner race). This provides direct evidence of an early inner race defect. Sidebands surrounding BPFI confirm repeated rolling element impacts at the defect. Tangential direction is highly sensitive for early-stage detection, making it suitable for long-term trend monitoring.

14 Fig. 7 - Spectrum Tangential (Low Range) – Pump Bearing #4 (Free End, 6321) Figure 8 shows a peak around 9.84× order (0.62 mm/s) indicates harmonic/sideband build-up. Although amplitudes are lower than axial/tangential, the radial spectrum confirms that defect energy is spreading. This spectrum acts as secondary evidence supporting inner race fault diagnosis. Fig. 8 - Spectrum Radial (Low Range) – Pump Bearing #4 (Free End, 6321) Figure 9 shows the physical condition of Pump Bearing #4 (type 6321) after disassembly, confirming the presence of a localized inner race spalling defect. The damage is characterized by a distinct pit and material flaking along the raceway, consistent with surface fatigue failure resulting from repeated rolling contact stress. This type of defect typically develops when localized subsurface cracks propagate to the surface under cyclic loading. The physical evidence aligns with the vibration analysis results, where the BPFI (Ball Pass Frequency of Inner Race) and its harmonics were dominant in both the low- and high-range spectra, indicating progressive inner race degradation. The visual confirmation validates the diagnostic conclusion obtained from the TWF and FFT analyses. Fig. 9 - The physical condition of Pump Bearing #4 (type 6321) after disassembly 5. Conclusion It is clear that the TWF recorded from Pump Bearing #4 corresponds to an inner race fault. The repeating impulses, the spacing equal to 1× shaft speed, and the evident amplitude modulation all support this diagnosis. Both low-range and high-range spectra provide complementary insights into bearing fault progression. The low-range spectra capture early-stage fault indicators with smaller amplitudes and sidebands near BPFI and BSF, while the high-range spectra emphasize advanced fault signatures with strong harmonics at 21×–25× orders. Together, they illustrate the evolution of the Pump Bearing #4 defect from early initiation to a critical failure stage. The low-range spectrum already contains BPFI-related content and impact sidebands, supporting an inner-race fault diagnosis. The subsequent dominance of 21×–25× harmonics in the high range reflects fault progression and resonance excitation. References [1] A. M. Al-Ghamd and D. Mba, \"A comparative experimental study on the use of acoustic emission and vibration analysis for bearing defect diagnosis,\" NDT & E Int., vol. 39, no. 6, pp. 419–428, Sep. 2006. doi: 10.1016/j.ndteint.2006.01.005. [2] Fluke Corporation, \"Fluke 810 vibration tester: Diagnostic technology white paper,\" Fluke Corporation, Everett, WA, 2010. [Online]. Available: https://www.fluke.com. [3] A. K. S. Jardine, D. Lin, and D. Banjevic, \"A review on machinery diagnostics and prognostics implementing condition-based maintenance,\" Mech. Syst. Signal Process., vol. 20, no. 7, pp. 1483–1510, Oct. 2006. [4] R. B. Randall and J. Antoni, \"Rolling element bearing diagnostics—A tutorial,\" Mech. Syst. Signal Process., vol. 25, no. 2, pp. 485–520, Feb. 2011. doi: 10.1016/j.ymssp.2010.07.017. [5] N. Tandon and A. Choudhury, \"A review of vibration and acoustic measurement methods for the detection of defects in rolling element bearings,\" Tribol. Int., vol. 32, no. 8, pp. 469–480, Aug. 1999. doi: 10.1016/S0301-679X (99)00077-8. [6] H. P. Bloch and F. K. Geitner, Machinery Component Maintenance and Repair, 3rd ed. Oxford, UK: Butterworth-Heinemann, 2011.Inner race pitting/spalling

15 Development of Waste Separation System with Microcontroller System Integration as Learning MaterialAbdul Mustaqim Bin Nordin1, *, Nurul Afiqah Bt Abdul Manas1, Darwinvellu A/L Shanmugam1, Muhammad Firdaus Bin Ishak1 1Department of Mechanical & Production Engineering Technology, Mechatronic Engineering Technology Division, Japan-Malaysia Technical Institute, 59, Lorong Perindustrian Bukit Minyak, 15, Kawasan Perindustrian Bukit Minyak, 14100, Simpang Ampat, Penang Malaysia *Corresponding Author Email: [email protected] 1. IntroductionThis project focuses on creating an educational prototype that demonstrates the use of automation and control systems in environmental applications. This system aims to classify and separate different types of waste, such as metal and nonmetal using sensors and microcontroller-based control logic. By integrating Internet of Things (IoT) and embedded system principles, the project enhances students’ understanding of real-world industrial automation processes while promoting sustainable practices. The design provides a hands-on learning platform for students in engineering and mechatronics fields to explore system integration, sensor calibration, and programming. Through this project, learners can experience how smart technology contributes to efficient waste management, environmental protection, and resource optimization. Ultimately, this project serves as both a practical teaching tool and an innovation model for modern automation education.1.1 Problem statement Most of the training stations focus on separating materials by color. This station, which is equipped with a conveyor, will separate materials according to the designated color and push them into a specific section by sliding. Based on observations, students lack knowledge about separating materials according to the type of material, whether metal or non-metal. In addition, existing training stations have not yet been combined with Internet of Things (IoT) elements. Most existing separation stations use Programmable Logic Controller (PLC) or Arduino to control the project system. Hence, the students are unable to clearly see the application of Internet of Things (IoT) on different types of material separation systems. Based on observations made on training stations in the automation lab, the following is a statement of the problem that can be formulated.  Lack of different types of learning stations in the automation lab  Students lack exposure to the process of detecting and separating metal and non-metal materials.  Lack of exposure to the Internet of Things (IoT) system. 1.2 ObjectivesThis project was developed to produce a training station to detect and separate metallic and non-metallic materials. There are three (3) objectives to be achieved at the end of this project. The objectives of this project are to:  Design a waste separation system for metallic and nonmetallic materials.  Produce a waste separation system based on the design that has been produced. Abstract: Modern technical education requires practical learning stations that reflect current industrial automation, focusing on sensor technology and network connectivity. This project addresses the shortage of specialized equipment at Japan-Malaysia Technical Institute (JMTI), where students demonstrated low exposure to both material sorting processes and Internet of Things (IoT) applications. The research aimed to design, develop, and program a functional, small-scale waste sorting system as a supplementary teaching tool for mechatronics modules. The prototype is a complex, integrated electromechanical system merging a mechanical structure, pneumatic actuators, and a microcontroller-based control system. The methodology utilized the Arduino Mega for real-time control and a NodeMCU ESP 32—implementing IoT via the Blynk platform. An inductive sensor was leveraged to successfully differentiate and sort metal items while infrared sensor was leveraged to successfully count the sorted part. Comprehensive testing verified the system’s reliability, though initial phases required rigorous troubleshooting for sensor malfunction and IoT connectivity issues. The project’s core educational utility was decisively validated. Post-training assessments showed a significant, uniform increase in student competency, moving from \"No knowledge\" to \"Has knowledge\" across core areas including system design, mechanical assembly, electrical wiring, and programming. Consequently, the final system provides a proven, functional, hands-on learning platform that robustly applies IoT concepts in an automation context. Future enhancements suggested including incorporating colour sensing and enabling remote ON/OFF control via Blynk. Keywords: Material sorting, Internet of Things (IoT), sensor, industrial automation, microcontroller.

16  Program a material separation system by applying the Internet of Things (IoT) system. 1.3 ScopeThis developed project has several limitations when carrying out its function as a waste separation station. This limitation is also known as the scope of the project, where the functionality of this project is limited to several aspects. The scope of the project identified is:  Only block-shaped iron materials can be detected and separated by this system.  This system can only detect and separate materials of the following sizes: o Minimum - Width: 11 cm, Length: 14 cm, Height: 5 cm o Maximum - Width: 16 cm, Length: 25 cm, Height: 10 cm  The detection and separation process can only be done on one material at a time. 2. Methodology Methodology is a systematic theoretical analysis of the methods used in a field of study. It consists of theoretical analysis, a body of methods and principles related to a branch of knowledge. It usually includes concepts such as theoretical models, steps set to ensure the best in the process of completing a project. 2.1 Project Design The design process begins with a three-dimensional (3D) sketch produced part by part mechanically and followed by assembly and detail drawings. The design process then continues by producing two-dimensional (2D) drawings showing top, front and side view sketches. Fig. 1 – 3D Drawing 2.2 Bill of Materials (BOM) Some of the materials used in the project as per listed below. 2.3 System Assembly Some of the assembly activities as per listed below. 2.4 Wiring Electrical wiring and pneumatic connections are done based on the circuits developed for the following categories.  Arduino Circuit  NodeMCU Circuit  Pneumatic Circuit

17 Fig. 2 – Arduino Circuit Fig. 3 – NodeMCU Circuit Fig. 4 – Pneumatic Circuit 3. Finding and AnalysisThe project report highlights a successful outcome in both the technical function and educational utility of the material sorting system. Comprehensive testing successfully addressed and resolved three key operational issues. Firstly, a connectivity failure on the Internet of Things (IoT) application via Blynk was resolved by simply resetting the mismatched passwords, resulting in acceptance. Secondly, the system's initial inability to identify waste material was permanently fixed by replacing the faulty sensor and adjusting its position, overriding the temporary measure of just adjusting the original sensor's location. Lastly, the problem of the lifter stalling midway in the chute was corrected by resetting the program delay, which was found to be too short, confirming the system's operational timing control. Beyond technical function, the project proved highly effective as a training tool in the Mechatronics workshop. Student evaluation clearly demonstrated that exposure to the project station successfully transferred practical knowledge across all tested areas. Before the training, students indicated they had no knowledge regarding the system's design, mechanical installation, electrical wiring (controlling the motor and sensor), and operational programming. After using the station for practical sessions, students confidently attested to having knowledge in every one of these core technical competencies. This confirms the system achieved its objective of supplementing existing learning stations and providing crucial exposure to metal/non-metal sorting and IoT concepts. 4. ConclusionThe project's discussion section addressed key developmental challenges, primarily stemming from the design's complexity due to the integration of mechanical, electrical, and pneumatic components. Further complications included the high risk of component damage from electrical wiring mistakes and the complexity of programming needed to coordinate multiple elements like motors and sensors. To mitigate these issues, the team proposed referring to existing designs for upfront clarity, having experienced personnel review electrical wiring before powering the system, and adopting a phased, incremental approach to coding guided by the project's flow chart. Ultimately, the project was deemed a success, fully achieving its initial objectives. The system was completed, tested, and demonstrated its ability to detect and sort metallic materials by successfully applying the Internet of Things (IoT) concept. Beyond its primary technical function, the project fulfilled its core purpose as an educational tool: it is now actively used as training material in the workshop, serving as a valuable supplementary learning station that successfully exposes students to different sensor types and the practical application of IoT principles. For future iterations, the developer suggested upgrading the sensor capability to detect colour and different sizes and enhancing the Blynk system to allow the project to be remotely turned ON/OFF via a smartphone. Acknowledgement The project team would like to express sincere appreciation to all individuals and organizations who have contributed to the successful completion of this project entitled Development of Waste Separation System with Microcontroller System Integration as Learning Material. Special gratitude is extended to Head of Mechatronics Division, for his invaluable guidance, professional advice, and continuous support throughout all stages of this project. The team also wishes to acknowledge the Department of Mechatronics Engineering, Japan-Malaysia Technical

18 Institute (JMTI), for providing the necessary facilities, resources, and technical assistance that enabled the successful implementation of this work. The team further extends appreciation to fellow students, laboratory staff, and all individuals who offered constructive feedback and cooperation during the project’s development. Lastly, heartfelt thanks are conveyed to our families for their unwavering encouragement, patience, and understanding, which have been instrumental to the completion of this endeavor. Appendix A: 2D Drawing – Top ViewAppendix B: 2D Drawing – Side ViewAppendix C: 2D Drawing – Front ViewAppendix D: 2D Drawing – Blynk Apps DisplayReferences [1] Rizki Abdullah, (17 Mac 2017), Kit pembelajaran pengasingan menggunakan conveyor. Retrived https://youtu.be/g3SusSTyqAK. [2] Nevon Projects, (12 Jun 29180), PLC Based Product Sorting Machine System. Retrieved https://youtu.be/UJl8JVOMCsw. [3] SysAdmin Adtec Melaka, (10 April 2019), kit pembelajaran pengasingan benda kerja mengikut warna. Retrived https://youtu.be/ WL7-eRvqm. [4] 1. Joel H. Spring, (1945), United Kingdom, The Sorting Machine National Educational Policy, Longman 1989, pg 100-119. [5] Official Gazette: (1993), United States. Closed Loop Conveyor Unit, pg 1-146. [6] Jozef Grochwicz: (1980), United Kingdom. Machines for Cleaning and Sorting Seeds, pg 1-116.

19 Integrated Leaked Gas Detector with Voice Alert and IoT System Muhammad Faris Adnan 1, Muhammad Muqri Rafiee 1, Muhammad Asnoor Abas 1, Mohd Hezri Abdullah 1, *, Shakirah Yatim 1, Irwan Zuraimi Md Naim 1, Azman Supa 1, Karolyn Anne Andrew Fabian 11Department of Electronic & Computer Engineering Technology, Electronic Engineering Technology Division, Japan-Malaysia Technical Institute, 59, Lorong Perindustrian Bukit Minyak, 15, Kawasan Perindustrian Bukit Minyak, 14100, Simpang Ampat, Penang, Malaysia*Corresponding AuthorEmail: [email protected] 1. Introduction Liquefied petroleum gas (LPG) usage has become the norm in every house nationwide in Malaysia for cooking purposes. Most of the houses use cooking gas, normally sold in a 14 kg gas cylinder. Centralised gas piping in Malaysia is normally used for premium condominiums. Thus, there is a need for a mechanism for detecting any gas leakage from the gas cylinders. The logistic process of the gas cylinder delivery, as well as the long storage of the gas cylinders in the warehouse, creates a high probability of gas leakage occurring before delivery to customers. Currently, there is no effective mechanism for detecting gas leakage from the gas cylinder. Generally, there are two main potential hazards due to gas leakage. The possibility of an explosion can happen if the mixture of LPG and the existing air is within the specification of an explosion, as well as having a fire source within the environment. Secondly, there is a possibility of suffocation whenever the oxygen concentration level has dropped due to the increased concentration of LPG in the surroundings. The lack of maintenance for the gas hose can trigger the LPG gas to leak. In addition, the leakage can also happen due to the usage of a lower-quality or an old gas hose. A quite dirty kitchen can also attract rats and insects, which can bite the gas hose. The usage of a gas hose made from transparent plastic can harden and break if used for quite a very long period of time, which can trigger gas leakage. Thus, it is imperative to detect any gas leakage, particularly in domestic households, due to the potential of the foreseen hazards. The project is initially conceived as a platform for developing a more comprehensive design in the future. A series of improvements in terms of the design and component selection can be done for the prototype in the near future. This paper is divided into 5 main sections, starting with the Introduction section. Section 2 is about an overview of the Integrated leak gas detector system. Section 3 explains the methodology of the design. Section 4 delves into the results and discussion about the design. Finally, Section V concludes the paper. 2. Overview of the Integrated Leaked Gas Detector System and its Structure The work is developed for integrating the microcontroller with 2 sensors, namely gas and weight sensors. The system also uses an alarm circuit and IoT concept via the Blynk app for notification of any gas leakage based on gas volume variation [1]. Specifically, the system uses an MQ6 gas sensor, an alarm circuit of ISD182O model, and a Load cell 50kg circuit as a weight sensor cum weight reader. Abstract:The Integrated leaked gas detector with voice alert and IoTs system is designed for detecting leaked gas in cooking gas cylinder for the Malaysian domestic household environment, since the cooking gas is purchased individually in a specified gas cylinder, weighing about 14 kg. The integrated leaked gas detector system uses a microcontroller, MQ-6 gas sensor, ISD1820 (voice recording module) as a triggered alarm, and a load cell sensor (maximum specification of 50 kg) as a weighing scale for detecting any leakage at the individual cooking gas cylinder. The Arduino ESP8266 NodeMCU microcontroller integrated with Wi-Fi isused as the main circuit controller, and the alarm circuit will be triggered once any gas leakage is detected. The Internet of Things (IoTs) concept has been embedded in the system by using “Blynk” apps for monitoring the gas volume in the individual gascylinder. The system has a better performance in terms of detecting the leakage gas at a maximum range of 30 cm by using the MQ-6 gas sensor with a response time of 0.16 seconds. Compared to the previous system whereby the detector system uses an MQ-2 gas sensor with a maximum range of 10 cm and a 0.30-second response time. Thus, the proposed integrated leaked gas detector has a better performance of about 33 % in terms of its detection range of leakage gas compared to the previous leakage gas detector system using the MQ-2 gas sensor. In addition, the response time of the former integrated gas detector system has improved response time of about 53.3 % faster compared to the latter gas detector system. All in all, the integrated gas detector system is very practical and beneficial as the first layer system for detecting any gas leakage in the domestic household environment in Malaysia.Keywords:Integrated Gas Detector, IoTs (Internet of Things), MQ-6 Gas Sensor, ESP8266 NodeMCU.

20 The kitchen is set to be the location for this system to be applied. Fig.1 shows an overview of the system being developed for detecting leaking gas. The work uses a microcontroller from the ESP model (Espressif Systems) integrated with built-in Wi-fi [2]. The system uses an alarm circuit that will be triggered automatically once any leaked gas is detected. The system is linked to an app known as Blynk for monitoring the volume of gas in the gas container. As a starting point, the system is only suitable for a gas cylinder with a maximum weight of 14 kg due to the constraint of the load cell specification that has been used. The physical structure of the prototype is shown in Fig. 2. The gas sensor is located on top of the enclosure to ensure that the detection of any leaked gas can be immediately traced. The load cell (weight sensor) is at the bottom for the weight, which represents the volume of the gas in the cylinder. The gas cylinder shall be put in the enclosure until the gas volume has depleted if there is any leakage of gas. The circuit box contains the microcontroller, voice recording module ISD1820, LCD display, and speaker unit. Fig. 1 - The structure of the system Fig. 2 - The physical structure of the system 3. Methodology Initially, the system is developed by designing using a design software known as Tinkercad. Fig. 3 shows the circuit in a Tinkercad layout environment. It is a browser-based electronic circuit simulator that supports Arduino microcontrollers as well as ATtiny chips. Fig.3 - The layout of the circuit in Tinkercad environment. Then, the design is fabricated and tested. The results are as in the following section. Fig.4 shows the block diagram mechanism of the system. Both inputs, namely gas and weight sensors, are used simultaneously. Thus, the leaked gas will be detected by the gas and weight sensor. The alarm system, i.e, the notification via the Blynk application and voice alarm, shall be activated [3][4]. The LCD will show the current status of gas inside the cylinder [5]. In that case, the potential hazard can be avoided since 2 types of sensors as well as 2 types of alarms are being used simultaneously. Fig.4 - The block diagram of the system mechanism. Fig. 5 is the structure of the voice alarm being used in the system. The microphone is for making a prerecorded voice before being implemented in the system. The voice will be output from its speaker unit (Not shown in Fig.5). Fig.5 - Alarm voice circuit Fig.6 shows the final layout of the prototype system. Most of the hardware components are put outside the enclosure, except the gas sensor (top of the enclosure) and the load cell (weight sensor at the bottom of the enclosure).

21 Fig.6 - The final layout of the prototype system in the circuit box Fig. 7 is the complete system with enclosure, gas, and weight sensors, as well as the circuit box. Fig.7 - The complete system of the prototype All the tests are conducted in several phases, beginning with mechanical checks during assembly and followed by electronics and system integration tests. 3. Results and Discussions This prototype, using an MQ6 gas sensor, indicates that it can identify the leaked gas from distances of up to 30 cm in as little as 0.16 seconds. The result is obtained in a controlled environment setup using high gas concentration (~1000 ppm LPG concentration) with optimal conditions such as room temperature of about 20 ̊C, 65% humidity, proper preheating for about 24 to 48 hours, and specific load resistance of about 20K Ohms for detecting LPG gas. The scenario represents the MQ6 sensor's maximum potential speed of initial detection instead of the average response time in a typical household environment. In comparison, for a controlled environment using a system having an MQ2 sensor, its performance can only achieve a maximum range of 10 cm with a 0.30-second response time. Thus, both scenarios are considered as benchmarks for both types of sensors. For the real scenario, this prototype has been tested using normal room conditions, gas leakage in a typical Malaysian household environment, and no specific load resistance for detecting leaked LPG gas. The MQ-6 and MQ2 gas sensor system response time increases as the distance from the gas source increases due to gas dispersion [6]. The optimal placement for the sensor, considering both response time and detection capability, is between 20 cm and 40 cm from the potential source of a leak. The performance result of the system using the MQ-6 and MQ-2 gas sensors is shown in Table 2. For the 0 cm (range between the leaked source and the gas sensor), the response time of both systems is slightly delayed due to the endothermic process of the liquid LPG vaporizing and absorbing heat from the sensor's heating element. The 20 cm range has the average response time slightly longer at about 1.01 and 0.97 seconds, respectively. This distance provides an optimum condition and fast detection point since the gas has sufficient duration for properly dissipating and reaching the sensor's detection surface effectively. The 40 cm range has a mean response time of about 1.33 and 1.31 seconds. Fortunately, both readings is still very fast and within the manufacturer's specified general response time of <10 seconds. Table 1 - Comparison of response time between the prototype using MQ-6 and MQ-2 gas sensors in real domestic household conditions. MQ-6 gas sensor system ( This work) MQ-2 gas sensor system Range (cm) Mean response time (s) Mean response time (s) 100 1.31 1.29 80 1.134 1.49 60 1.39 1.38 40 1.33 1.31 30 1.01 1.10 20 1.01 0.97 10 0.84 0.89 0 0.94 1.13 The 60 cm range and beyond has a fluctuating response time for both types of sensor systems. The scenario happens due to the concentration of the gas decreases significantly as the distance increases. Thus, the situation makes the leaked gas detection system unreliable. Thus, it causes the sensor to be unable to trigger an alarm. The specification of the weight for the gas, including its cylinder, converted into a percentage, is as in Table 2. The specification has been developed after thorough tests, whereby the weight of several gas cylinders is measured individually. Fig.8 shows that the system operates well within the specified specifications in Table 2. The percentage value and weight are displayed on the LCD for reference, and meanwhile, the voice alarm will be triggered whenever any leakage gas occurs. The IoTs concept using the Blynk application has also operated well. The notification appears on the smartphone as in Fig.9. In short, the whole alarm system and notification via LCD and Blynk application are activated within a short span of time.

22 Table 2 - The specification of estimated gas volume in weight and percentage. Weight ( kg ) Percentage ( % ) 30.5-30.0 29.1 – 28.6 27.7 – 27.2 26.3 – 25.8 24.9 – 24.4 23.5 – 23.0 22.1 – 21.6 20.7 – 20.2 19.3 – 18.8 17.9 – 17.4 16.5 and below 100 90 80 70 60 50 40 30 20 10 0 Fig.8 - The output on the LCD from the prototype system in a real domestic household condition. 5. Conclusion The Integrated leaked gas detection system with voice alert and IoT system prototype has demonstrated the feasibility of creating a functional leaked gas detection alarm system with the IoTs concept. The prototype has shown the capability to respond to any gas leakage within less than 2 seconds, and the notification as well as alarm can be triggered within less than 10 seconds. Thus, the prototype system has proved the future capability to be applied in a wider scope, such as in a logistics gas store or gas factory. Acknowledgement The authors wish to thank the Japan-Malaysia Technical Institute and particularly the Electronics Department for the provided expertise and the use of equipment. References [1] M. A. Bid, “Monitoring the gas cylinder level and gas seepage detection through IoT,” Int. J. Emerg. Technol. Eng. Res. (IJETER), vol. 6, no. 4, pp. 233–236, 2018. [2] N. V. Rohith et al., “Smart LPG gas level detection and safety system using IoT,” Int. J. Eng. Res. Technol. (IJERT), vol. 8, no. 13, pp. 215–218, 2020. [3] Z. S. Zainal and A. Masek, “Research and innovation in technical and vocational education and training,” Research and Innovation in Technical and Vocational Education and Training, vol. 4, no. 2, pp. 114–120, 2024. [4] R. M. Yusuf and A. P. Wahyu, “Internet of Things-based gas leak detection with alerts via SMS and Blynk app,” Jurnal dan Penelitian Teknik Informatika, vol. 7, no. 3, pp. 812–816, 2022. [5] B. S. Syeda and P. Ch. R., “Gas leakage detection and alerting system using Arduino Uno,” Global Journal of Engineering and Technology Advances, vol. 5, no. 3, pp. 29–35, 2020. [6] E. Fatkiyah et al., “Early detection of leaks on gas cylinders using Arduino-based MQ-6 sensors,” J. Phys.: Conf. Ser., vol. 1413, p. 012030, 2019.Fig.9 - The system’s notification displayed on the smartphone using the Blynk application.

23 Remote Controlled Bionic Hand Muhammad Idham Mohd Sebi 1, Mohd Azha Mohamad Salleh 1, Mohd Hezri Abdullah 1, *, Shakirah Yatim 1, Irwan Zuraimi Md Naim 1 ,, Azman Supa 1, Karyna Arania Abdullah 1 1Department of Electronic & Computer Engineering Technology, Electronic Engineering Technology Division, Japan-Malaysia Technical Institute, 59, Lorong Perindustrian Bukit Minyak, 15, Kawasan Perindustrian Bukit Minyak, 14100, Simpang Ampat, Penang, Malaysia*Corresponding Author Email: [email protected] 1. Introduction Industrial application of the Robotic hand has become the norm in an industrial environment. However, the transmutation and miniaturization of robotic hands for imitating gestures or movement of humans is a fastdeveloping active field in the research of robotics applications, particularly in the education field [1]. The concept of a robotic hand for imitating the gesture or movement of humans has gained popularity since the French sculptor Gaël Langevin developed his open-sourced, 3D printed humanoid robot starting in 2012 [2],[9]. The humanoid robot is known as InMoov. The project is initially conceived as a platform for learning robotics, but it has expanded into many iterations in many fields, such as a cheap stand-in for prosthetics, a technology demonstrator for machine vision, and HumanMachine Interaction. In the case of the bionic hand, it operates through a series of servomotors strategically placed along the arm to replicate natural human arm and hand movements. This paper is divided into 5 main sections, starting with the Introduction section. Section 2 is about an overview of the remotely controlled bionic Hand design. Section 3 explains the methodology of the design. Section 4 delves into the results and discussion about the design. Finally, Section V concludes the paper. 2. Overview of the Remote Controlled Bionic Hand and its Structure The remote-controlled bionic hand operates by replicating human hand movements through real-time data acquisition, wireless communication, and precise actuation. There are two main components in the system, namely transmitter nodes and a receiver node. Both utilize ESP32 microcontrollers. For the transmitter, hand movements are captured using an Inertial Measurement Unit (IMU) and flex sensors. The IMU used, namely the MPU6050 [3], performs sensor fusion on data from the accelerometer, gyroscope, and magnetometer. This fused orientation and motion data is accessed by the ESP32 via the I²C communication protocol [4]. Simultaneously, flex sensors detect the bending of individual fingers, outputting analog signals that are read through the ESP32’s Analog-to-Digital Converter (ADC). The ESP32 processes this data and formats it for wireless transmission. The system uses ESP-NOW, a peer-to-peer protocol based on Wi-Fi (IEEE 802.11b/g/n) that enables fast, low-latency data transfer without requiring a router or internet connection. The processed motion and gesture data is transmitted wirelessly from the transmitter ESP32 to the receiver ESP32 as a means for communication between nodes. On the receiver side, the ESP32 decodes the incoming quaternion and flex sensor data. Later, it converts the interpreted motion into corresponding servo angles and generates Pulse Width Modulation (PWM) signals for Abstract:This paper focuses on a remote-controlled bionic hand designed to serve as a teaching aid in technical colleges as well as in the robotics industry. The remote-controlled bionic hand designed in this work is aimed at being used as teaching material consisting of various technology knowledge from 3D printing, electronics, programming, wireless communication until robotics field. The system uses the concept of human-machine and sensor-based interaction. The system is operated using data obtained from Inertial Measurement Units (IMU) and Flex sensors, which capture elbow and finger movements in real time. The ESP32 processes this data and formats it for wireless transmission via the ESP-NOW protocol to another ESP32 operating as a receiver node. The ESP32 decodes the incoming quaternion and flex sensor data and converts the interpreted motion into corresponding servo angles. Then, the generated Pulse Width Modulation (PWM) signals from the second ESP32 are sent to control the bionic hand's joints. Each servomotor is mapped to a specific joint or finger, allowing for real-time replication of the user's arm and hand movements. Thus, the movement of the human elbow, forearm, or fingers will be mimicked by the remotely controlled bionic hand in real time. In this work, the servo motor at a set angle of 10 degrees represents elbow movement at 0 degrees (or no movement), while the maximum angle of the servo motor, set at 60 degrees, produces a maximum elbow movement of 90 degrees. On the other hand, for finger angle movement, the servo motor angle set at 80 degrees can represent the fingers' movement at 0 degrees (i.e, no movement) while the servo motor angle set at 180 degrees represents the finger movement at 90 degrees (maximum movement of the fingers), creating the key features of human-like movement replication. Lastly, the system demonstrates a viable approach in teaching embedded systems, wireless communication, as well as applications in the robotics field within the technical and vocational education and training (TVET) environment. Keywords: Bionic Hand, IMU Sensor (or Inertial Measurement Unit), ESP-NOW Protocol, Servo motor, Pulse Width Modulation

24 controlling the bionic hand's joints [5]. Each servomotor is mapped to a specific joint or finger, allowing for real-time replication of the user's arm and hand movements. For example, when the user bends their index finger, the flex sensor value is received by the ESP32, which in turn actuates the servo responsible for that finger [6]. Fig.1 and Fig.2 show the degree of freedom for fingers and the elbow, respectively. All the fingers’ movements are based on the forearm section. The forearm section consists of six servos, whereby five of these servos control the individual fingers. Each finger is actuated by a tendon-like mechanism, where a string is connected from the servo to the fingertip. When the servo rotates, it pulls the string, causing the finger to bend, effectively mimicking the flexion movement of a human hand. The sixth servo in the forearm controls the wrist rotation, using a series of spur gears to rotate the hand along the axis of the arm [7]. A single servo is used to control bending and extension at the elbow joint. The condition is achieved through a pistongear mechanism that provides the necessary torque and angular movement to simulate the elbow’s natural motion. Fig. 1 - Degree of freedom Fig.2 - Degree of freedom (DOF) of the human fingers. (DOF) of the human elbow [8]. [8]. 3. Methodology 3D printing or also known as additive manufacturing or fused deposition modeling, is used as the fabrication method for the mechanical components of the bionic arm. The mechanical parts are sourced from the InMoov project. They are compartmentalized into hand, forearm, elbow, and shoulder sections to ease print management and avoid confusion between almost similar parts. The InMoov project is open-sourced, thus enabling users to modify and provide further improvements to the project [9]. Fig.3 illustrates the operation of the prototype system, including the transmitter and its receiver nodes. All types of communication between different blocks have been displayed. Fig. 3 Block diagram between the transmitter and receiver nodes of the remotely controlled bionic hand system. [9] The operation principle of the prototype is shown in Figs. 9 and 10. The prototype operates by replicating human hand movements through real-time data acquisition, wireless communication, and precise actuation. The system consists of two main components: transmitter nodes and a receiver node, both utilizing ESP32 microcontrollers. The transmitter nodes (Node 1 to Node 5) can be worn on the human body, while Node 5 is at the bionic hand. Node 1, shown in Figure 4, transmits quaternion data representing the orientation of the upper arm. Node 2 transmits quaternion data corresponding to the forearm, which is used along with the raw quaternion from node 1 to determine elbow motion. Node 3 transmits quaternion data along with five analog readings from flex sensors, which are used to estimate the bending of individual fingers. Node 4 transmits quaternion data as the global reference for node 5 to calculate the relative orientation of each node. It is also equipped with a physical reset button that, when pressed, sends a broadcast message (RESET_CMD) to reinitialize the DMP on Nodes 1 to 4. Node 5 serves as the central receiver and controller. It processes quaternion and flex sensor data, calculates joint angles, drives the appropriate servomotors, and transmits a JSON-formatted telemetry packet including quaternion values, raw flex sensor data, and current servo angles to a local HTTP server for real-time monitoring. Fig. 4 - Node 1 ( Worn on the upper arm of a human) Fig 5 - Node 2 (Worn on the forearm of a human)

25 Fig 6 - Node 3 (Worn at the hand of a human) Fig.7 - Node 4 (Worn at waist) is a reference node, (Has a reset button to reinitialize MPU6050 DMP when drift accumulation became unacceptable). Fig.8 - Fully assembled bionic hand Figure 9 illustrates the flowchart of the operational flow of the ESP32 transmitter nodes (Node 1 to Node 4). Upon power-up, each node initializes the ESP32 and begins reading sensor data. Node 1 and Node 2 read IMU data (quaternion), while Node 3 reads five flex sensor values for finger movement detection. The collected data is then processed by the ESP32 microcontroller and transmitted via the ESP-NOW protocol to the receiver node (Node 5). Node 4 includes a reset button that, when pressed, sends a broadcast reset command to all transmitter nodes, reinitializing their IMUs. The system loops continuously until powered off. Fig.9 - Transmitting signal flowchart Fig.10 - Receiving signal flowchart Testing was conducted in several phases, beginning with mechanical checks during assembly and followed by electronics verification and system integration tests. 3. Results and Discussions The prototype is built to replicate basic human arm and hand functions with reasonable fidelity and responsiveness by leveraging open-source resources from the InMoov project, 3D printing technology, and readily available components such as the ESP32 microcontroller, servo motors, and MPU6050 sensor [9],[10]. The mechanical assembly is based on the use of PLA filament. The structure demonstrates acceptable structural integrity and dimensional accuracy. On the receiving node, the bionic hand system uses servomotors to drive individual joints and fingers. Functional testing shows that, while full anatomical motion is not achieved, the key features of human-like movement are effectively replicated. The IMU used, namely MPU6050, delivers consistent and usable quaternion data, enabling gesture-based control with an acceptable level of accuracy. Wireless communication via ESP-NOW performs well within the test environment, providing a low-latency, stable data link between the control unit and the robotic arm. Although formal range and latency testing is not performed, the communication protocol demonstrated sufficient reliability for the intended application. Table 1 shows the performance of the servo motor angle and the degree of elbow and finger motion. There is a strong correlation between the angles of servo motors and the

26 movement of joints. Initially, the servo motor’s angle is set at 10 ̊ (i.e, no movement) and can extend to 60 ̊ (maximum movement of the elbow equivalent to 90 ̊ of movement). On the other hand, the relationship of the servo motor’s angle and fingers’ movement starts at the servo motor with an angle of 80 ̊ (i.e, no movement of fingers) until the servo‘s angle of 180 ̊ (maximum movement of fingers at its maximum angle of 90 ̊ ). The system can be improved by controlling the servo’s rotation angle in order to ensure the robot’s elbow and fingers work with finer motion. The servo’s rotation angle can be modified by adjusting the pulse width modulation (PWM) signal or physically modifying the servo's internal mechanics.Alternatively, stepper motors can be used to replace the servo motors in the bionic arm system, particularly for less demanding tasks or lighter loads. Unfortunately, the usage of stepper motors comes with a trade-off in terms of speed, torque, and control. Generally, stepper motors are good for applications with repeatable or slow movements. However, in general, servo motors are better for high-torque, high-speed applications, whereby precise positional feedback is very important. Table -1 Performance of the servo motor angle and the degree of elbow and finger motion Servo motor angle ( degree) Elbow angle (degree) Fingers angle (degree) 0 NA NA 10 20 0 15 NA NA 40 60 80 100 120 140 160 180 75 90 NA NA NA NA NA NA NA NA 0 20 45 65 70 90 Note: -NA is not applicable The applications of 5 different nodes, as in Fig.4 to Fig.8, blend well with each other, which contributes to a complete system. The transmission of a signal via the ESP-NOW protocol from the ESP32 at the transmitter node to the ESP32 at the receiving node works well at a range of about 1.5 m. The transmission of the PWM signal from the ESP32 at the receiving node to the servo motors at the fingers and elbow does make the expected motions. Thus, the remote-controlled bionic arm system has proved the future capability of the system to be applied to a more complex robot arm design with a wider scope. 5. Conclusion The Remote Controlled Bionic Hand prototype has demonstrated the feasibility of creating a functional, gesturecontrolled bionic hand using low-cost, open-source tools and components. The prototype has shown the capability to mimic the degrees of freedom (DoF) of the human hand successfully. Thus, it can serve as an alternative teaching aid for demonstrating embedded systems, sensor integration, and robotic control at any training institution worldwide. Acknowledgement The authors wish to thank the Japan-Malaysia Technical Institute and particularly the Electronics Department for the provided expertise and the use of equipment. References [1] Y. Zhang and L. Zhang, \"Research on Education Robot Control System Based on ESP32,\" Journal of Education and Educational Research, vol. 7, no. 2, pp. 299-302, 2024. [2] L. Paral, A. Sari, and M. Esen, \"Design of a 3D Printed Open Source Humanoid Robot,\" Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, vol. 11, no. 2, pp. 411-420, 2022. [3] J. Huang, \"Design of Angle Detection System Based on MPU6050,\" in Advances in Computer Science Research (ACSR), vol. 73, pp. 6-8, 2017. [4] L. Zheng and M. Chen, \"The Design and Implementation of a Robotic Arm Digital Twin System Based on ESP32,\" International Journal of Advanced AI Applications, vol. 1, no. 6, pp. 46-61, 2025. [5] A. B. Buji, Y. P. Mshelia, A. G. Ibrahim, and M. A. Sarki, \"Model Design, Simulation, and Control of a Robotic Arm using PIC 16F877A Microcontroller,\" 2019. [6] M. H. Shaife and R. Hussain, \"The Development of an Elementary Robot Arm Using the Educational Simulation Tool,\" Journal of Engineering Technology, vol. 9, no. 1, pp. 73-76, 2021. [7] G. Katal, S. Gupta, and S. Kakkar, \"Design and operation of synchronized robotic arm,\" International Journal of Research in Engineering and Technology, vol. 2, no. 8, pp. 297-301, 2013. [8] F. Paulsen and J. Waschke, Sobotta Atlas of Human Anatomy: Vol. 1. General Anatomy and Musculoskeletal System, 15th ed. Amsterdam, Netherlands: Elsevier, 2011. [9] G. Langevin, \"Build Yours,\" InMoov. Accessed: Nov. 28, 2024. [Online]. Available: https://inmoov.fr/buildyours/ [10] K. M. Leong, T. S. Tan, M. T. F. Thye, J. Sheikh, B. S. Chan, D. Nordin, and J. H. Tan, \"Hand Gesture Design Using Bionic Prosthetic Hand,\" Journal of Human Centered Technology, vol. 2, no. 1, pp. 43-50, 2023.

27 Graphical User Interface (GUI) - Based OCR Application for Text RecognitionMuhammad Adam Danish bin Mohd Sazli1, Hakeil Shaquille bin Hashrulliza1, Nur Fatihah Aidiel binti Mohammad Fithry1, Ahmad bin Md Nizam1,*, Mohamad Luqman bin Mohamad Lotfi1, Wan Rohanina binti Wan Ibrahim1, Nor Yusuhanna binti Mohd Yusoff1, Hoo Seng Chun1, Zuraini binti Ghazali1, Ayub bin Mohd Rasidi1 1Department of Electronic & Computer Engineering Technology, Computer Engineering Technology Division, Japan-Malaysia Technical Institute (JMTi), 59, Lorong Perindustrian Bukit Minyak 15, Taman Perindustrian Bukit Minyak, 14100 Simpang Ampat, Penang, Malaysia *Corresponding Author Email: [email protected] 1. Introduction The advancement of digital technology has increased the demand for accurate and efficient text extraction from images. This project focuses on developing a Graphical User Interface (GUI)-based Optical Character Recognition (OCR) system using PaddleOCR to simplify and enhance the text recognition process. The main objectives are to improve OCR accuracy and reliability for distorted or unclear images, increase recognition confidence through model optimization, and organize recognized text results in a structured table for easier validation and analysis. This OCR system is developed to address the challenges users face in converting car plate text from images into digital form, particularly in cases where images are unclear, distorted, or complex. By incorporating a GUI, the system provides an intuitive interface, real-time feedback, and improved control over the recognition process, ultimately enhancing both user experience and text extraction performance. 2. Related Work Recent developments in OCR technology have introduced several open-source frameworks such as Tesseract [1], EasyOCR [2], and PaddleOCR [3], each offering distinct capabilities. Tesseract, developed by Hewlett-Packard and later maintained by Google, remains one of the most widely used OCR engines due to its stability and broad language support. However, studies have shown that its performance decreases when dealing with complex layouts, handwritten text, or low-quality images. EasyOCR, a lightweight PyTorchbased framework by Jaided AI, supports over 80 languages and provides quick implementation, yet it sometimes struggles with text alignment and font variations, limiting its consistency across diverse image inputs. Among these frameworks, PaddleOCR has emerged as a leading solution due to its high-accuracy convolutional recurrent architecture and flexible, modular design. Research indicates that PaddleOCR performs exceptionally well on multilingual and scene text datasets, especially when combined with advanced preprocessing techniques. Therefore, this study adopts PaddleOCR as the main OCR engine, enhancing it with custom preprocessing steps, a GUI for easier interaction, and visualization modules for real-time result interpretation. These improvements aim to optimize both system accuracy and user accessibility, ensuring a more efficient and reliable OCR application. 3. Methodology This section describes the methodology used to develop and evaluate the proposed system, illustrated through three main flowcharts. The first flowchart as shown in Figure 1 outlines the model development process, including validation, testing, data collection, and evaluation based on confidence and accuracy. The second flowchart as illustrated in Figure 2 explains the GUI design phase, which involves adding key features such as image and folder input, navigation controls, output validation, testing, and final execution. The third flowchart in Figure 3 shows the overall system workflow, combining GUI and OCR processes—from image input and preprocessing to text extraction, data validation, and output export. Overall, these flowcharts represent a structured and systematic approach to achieving the research objectives. Abstract: A Graphical User Interface (GUI)-based Optical Character Recognition (OCR) application was developed to automate the extraction and digitalization of text from images. The system integrates a GUI with an OCR model to enhance usability and efficiency. Implemented using Python, PaddleOCR, and Visual Studio Code (VSC), the application accurately converts imagebased text into digital form. Experimental evaluations demonstrated an accuracy exceeding 90%, underscoring the system’s reliability and performance. The results highlight the potential of GUI-integrated OCR systems as effective tools for rapid, accurate, and user-friendly car plate text recognition.Keywords: PaddleOCR, Optical Character Recognition, License Plate Recognition, GUI, Image Processing

28 The flowchart in Figure 1 illustrates the overall research process, beginning with the development of the model and ending with its evaluation. First, a model is created or selected based on the research objectives, representing the system or algorithm to be tested. The model then undergoes a validation process to ensure that it functions correctly and produces reliable outputs before being applied to actual data. After validation, a test data set is used to assess the model’s performance under real or simulated conditions. The results generated from these tests are collected for further analysis. The next stage encompasses the evaluation of two key aspects, namely confidence and accuracy Confidence measures the reliability or certainty of the model’s predictions, while accuracy determines how closely those predictions match the actual results. Both factors are then analyzed together in the evaluation stage to determine the overall effectiveness and performance of the model. This structured flow ensures that the research systematically validates, tests, and evaluates the model. Fig. 1 - Research Flowchart of OCR Development The flowchart in Figure 2 illustrates the development process of the system’s GUI, beginning from the design phase and concluding with the final execution. Fig. 2 - Flowchart of GUI Development The process starts by designing and implementing several essential features to enhance user interaction and functionality. During the design phase, features are added to allow users to insert images into the system, followed by another feature that enables the input of entire folders for batch processing. Additional components are included to improve usability, such as “Next” and “Back” buttons for smooth navigation between different sections of the interface. The system is further enhanced with a function to tabulate and export the output data, ensuring that users can easily manage and store their results. Another feature is also implemented to allow users to validate the generated output, ensuring that the information produced by the system is accurate and reliable. Once all these features are integrated, the system undergoes a testing phase to identify and correct any issues, ensuring optimal performance. Finally, after successful testing and verification, the project proceeds to the final execution stage, marking the completion of the GUI development process. This structured workflow ensures that all necessary features are systematically designed, tested, and executed to produce a functional and user-friendly system. The flowchart in Figure 3 illustrates the complete workflow of the OCR system, from input selection through preprocessing, data extraction, and final output validation.

29 Fig. 3 - Scanned Data Organized in Table Format The process begins with the Start node, leading to the Menu, where users are prompted to choose an input method within the GUI. Users can either input a single image, upload a folder containing multiple images, or use a connected camera to capture images directly. If the user selects the folder option, the system automatically processes every image within that folder. After input selection, the images proceed to the preprocessing phase, where each image undergoes OpenCV (cv2) preprocessing [4] to enhance quality, remove noise, and prepare it for text extraction. Once preprocessing is complete, the OCR model extracts text content from the images. The extracted data is then filtered through pre-approved formats to ensure that only valid and structured information is retained. The filtered and organized data is subsequently displayed in the GUI for user review. Following this, the process enters the output stage, where users can access the output validation menu. At this point, the system allows the user to verify the accuracy of the extracted information. If the output is validated as true, it is tabulated and saved in an Excel file as correct data. Conversely, if the data is validated as false, it is still exported to the Excel file but categorized separately as incorrect data for further review. The process concludes once all data has been validated and stored accordingly. Overall, this flowchart demonstrates a well-structured sequence for image input, OCR preprocessing, text extraction, and validation, ensuring that all outputs are systematically processed and organized for accurate and efficient data management.3.1 Otsu’s Threshold and Accuracy In this project, several equations were implemented to strengthen the accuracy and reliability of the OCR system. Otsu’s Threshold [5], is calculated using equations (1) and (2). The objective is to find the optimal threshold value, T, for separating text from the background during the image preprocessing stage. This method minimizes intra-class variance and maximizes between-class variance, ensuring that only clear and well-defined text regions are processed by the OCR model. Equation (3) involved the accuracy, A, calculation. By combining Otsu’s Threshold [6] with the accuracy calculation, the system achieved improved text recognition results, and enhanced reliability under various image lighting and distortion conditions. ?௪ଶ(?) = ?଴(?)?଴ଶ + ?ଵ(?)?ଵଶ(?) (1)?௕ଶ(?) = ?଴(?)?ଵ(?)(?଴(?) − ?ଵ(?))ଶ (2) ?(%) = ே௨௠௕௘௥ ௢௙ ௖௢௥௥௘௖௧௟௬ ௜ௗ௘௡௧௜௙௜௘ௗ ௜௠௔௚௘௦்௢௧௔௟ ௡௨௠௕௘௥ ௢௙ ௜௡௣௨௧ ௜௠௔௚௘௦ × 100 % (3)3.2 Preliminary OCR Model Before developing the complete OCR system with a GUI, several initial models were tested to assess and improve the accuracy, reliability, and confidence level of PaddleOCR in recognizing text from distorted or unclear images. A total of three models [7,8,9] were implemented throughout the testing phase, with each model designed to evaluate different processing approaches and performance outcomes as depicted in Figure 4. The latest and most effective approach, referred to as Model 3, represents the final optimized process that integrates pre-processing, PaddleOCR, and GUI for accurate, reliable, and structured text recognition and analysis. Fig. 4 - Processes and Sources in Related Optical Character Recognition (OCR) Models

30 3.3 Model 1: Direct OCR without Pre-processing As shown in Figure 5 for the first model, the image input was directly processed by the OCR model without any preprocessing or graphical interface. However, this approach revealed several limitations, such as lower recognition accuracy and reduced reliability when dealing with distorted or unclear images. The absence of enhancement and validation steps also led to frequent misread characters and unstructured output, making data analysis difficult. These issues highlighted the need to improve OCR accuracy and reliability through better image enhancement, optimize the PaddleOCR model for more precise text extraction, and develop a structured GUI for clearer result presentation.Fig. 5 - Flowchart of OCR Process without Preprocessing and GUI3.4 Model 2: OCR with Pre-processing (without GUI) In the second model as in Figure 6, the image input was processed through a pre-processing pipeline before being analyzed by the OCR model. Several enhancement techniques such as image resizing, noise reduction, contrast adjustment, and binarization were applied using the OpenCV library to improve image quality and readability. This process significantly enhanced OCR performance by reducing recognition errors, especially for blurred or low-contrast images. However, since the system did not include a validation mechanism and GUI, the recognized text still lacked structured presentation. These limitations emphasized the importance of not only improving OCR accuracy and confidence but also developing a GUI to organize and validate results in a clear, structured format. Fig. 6 - Flowchart of OCR Process with Pre-processing and without GUI 4. Results and Discussion Based on the evaluation conducted on a dataset of 100 images divided into validation, and testing sets, the third model demonstrated the best overall performance compared to the first and second model. This model integrated preprocessing, the PaddleOCR model, and a GUI, resulting in more accurate and structured text recognition. It showed strong performance across various testing conditions, including different angles, distances, and lighting environments, where the OCR system maintained high accuracy and recognition stability. The inclusion of preprocessing significantly enhanced the system’s ability to recognize and validate text even under these challenging conditions, reducing distortion effects and improving clarity. Optimization of the PaddleOCR model also contributed to higher recognition confidence by minimizing misread characters and ensuring consistent output. Additionally, the GUI effectively organized and displayed recognized text results in a clear table format, allowing users to validate and analyze data efficiently. The visual outcomes of these tests, covering variations in angle, distance, and lighting, are presented in Figure 7, Figure 8 and Figure 9. These figures illustrate how the third model consistently produced reliable and accurate text recognition results across different image conditions. Overall, this model successfully improves the OCR accuracy and recognition confidence, in addition to providing a structured, user-friendly data presentation system.

31 Fig. 7 - The subject in the image with angle variation Fig. 8 - The subject in the image with distance variation Fig. 9 - The subject in the image with lighting variation Table 1 summarizes the results of testing data used to evaluate the performance of the model. Each entry corresponds to an image processed by the OCR model, showing its confidence level in recognizing the plate text, the detected raw output, and the ground truth (GT), which is the actual plate text. The “GT is Correct” column indicates whether the OCR output matched the expected ground truth. As shown, all samples were correctly identified, demonstrating that the OCR program performed with 100% accuracy on this test dataset. The confidence scores, ranging between 0.90 and 1.00, indicate strong reliability and consistency in the model’s text recognition capability, even when minor variations such as spacing or formatting occurred in the input images. Table 1 - Scanned Data Organized in Table Format File-name Max Confid-ence Detected Plates Raw Text Ground Truth GT is Correct Image 23.png 1 VBR 9355 VBR 9355 Yes L2VXL.png 0.92 L2VXL L2 VXL Yes WG23EDR.png 0.98 WG23EDR WG23EDR Yes WN24FRX.png 1 WN24 FRX WN24 FRX Yes OZ7OZEV.jpg 0.9 OZ7O ZEV OZZO ZEV Yes 3679FLS.jpg 0.99 3579FLS 3579FLS Yes A2124.jpg 0.98 A2124 A2124 Yes AEQ5615.jpg 0.99 AEQ5615 AEQ5615 Yes B123WLG.jpg 0.94 B123 WLG B 123 WLG Yes Table 2 compares the three proposed models based on three main criteria: Recognition Accuracy, Confidence in Text Recognition, and Data Presentation and Validation. As shown, Model 1 demonstrates the weakest performance, achieving 0% accuracy and confidence due to the absence of a proper preprocessing or recognition framework, and it also lacks a data presentation interface, making the output unstructured and difficult to validate. Model 2, on the other hand, achieves 97% recognition accuracy and confidence, indicating significant improvement in text recognition reliability; however, its data validation remains limited as the results are only displayed through the console, offering minimal support for structured verification. Model 3 achieves the best overall performance with 97% accuracy and confidence, while also integrating a GUI that presents recognized text, original images, and confidence values in a clear, tabular format. This allows for easier comparison, validation, and exporting of data, providing a more user-friendly and efficient system for text recognition and verification.Table 2 - Comparison and Performance Evaluation Criteria Model 1 Model 2 Model 3 Recognition Accuracy 0% 97% 97%Confidence in Text Recognition 0% 97% 97%Data Presentation and Validation No data presentation interfacePartial validation through console displayFully integrated GUI presentation and validation

32 5. Conclusion The development of a GUI based OCR system using PaddleOCR has successfully achieved the main objectives of improving recognition accuracy, enhancing reliability, and providing structured and validated text results. Through three testing models, the final optimized approach (Model 3), which integrates image pre-processing, PaddleOCR model optimization, and a Tkinter-based GUI, produced the best overall performance. The use of OpenCV techniques such as noise reduction, contrast adjustment, and binarization effectively improved image clarity and allowed more accurate recognition of distorted or unclear text. Meanwhile, optimization of the PaddleOCR model increased recognition confidence by minimizing misread characters and providing confidence scores for each text segment. The implementation of a structured GUI interface enabled clear data visualization, displaying original images, recognized text, and confidence values in an organized table format, which simplified validation and analysis. Overall, the system provides a fast, accurate, and user-friendly OCR solution that transforms image-based text into reliable digital data. Future enhancements may include real-time recognition, database integration, and multi-language support to expand the system’s functionality and application scope. References [1] Smith, R. (2007). An overview of the Tesseract OCR engine. Proceedings of the Ninth International Conference on Document Analysis and Recognition (ICDAR 2007), Vol. 2, 629–633. IEEE. ISBN 978-0-7695-2822-9. Retrieved from https://doi.org/10.1109/ICDAR.2007.4376991[2] JaidedAI Research. (2020). EasyOCR: Ready-to-use OCR with 80+ supported languages. GitHub Repository. Retrieved from https://github.com/JaidedAI/EasyOCR[3] Du, Y., Xu, X., Wang, Y., & Baidu PaddlePaddle Team. (2020). PaddleOCR: An open-source OCR system based on deep learning. Proceedings of the PaddlePaddle Open Source Series. Retrieved from https://github.com/PaddlePaddle/PaddleOCR[4] Bradski, G. (2000). The OpenCV Library. Dr. Dobb’s Journal of Software Tools. Retrieved from https://opencv.org/[5] Nobuyuki Otsu, A Threshold Selection Method from Gray-Level Histograms, Published in IEE Transactions on Systems, Man, and Cybernetics, Vol. 9, No. 1, 1979, pp. 62-66. Retrieved fromhttps://doi.org/10.1109/TSMC.1979.4310076) [6] Didit Meidi, Yana Aditia Gerhana, Aldy Rialdy Atmadja, Popon Dauni, Implementation of OCR (Optical Character Recognition) Using Otsu Threshold Method for Detecting Tajweed Qur'an, EAI, 2021, [5-7]. Retrieved from eudl.eu/doi/10.4108/eai.11-7-2019.2298089[7] Irsyad, M. S., Che Embi, Z., & Ghauth, K. I. B. (2024). Assessing the Efficiency of Deep Learning Methods for Automated Vehicle Registration Recognition for University Entrance. Journal of Informatics and Web Engineering (JIWE), 3(2), 57‑69. Retrieved fromhttps://doi.org/10.33093/jiwe.2024.3.2.4 [8] Randika, A., Ray, N., Xiao, X., & Latimer, A. (2021). Unknown-Box Approximation to Improve Optical Character Recognition Performance. Lecture Notes in Computer Science (LNCS), Document Analysis and Recognition – ICDAR 2021, 12821, 481–496. Retrieved from https://doi.org/10.1007/978-3-030-86549-8_31[9] Patil, A., Gaikwad, P., Bhise, P., & Rathod, M. (2023). Automatic number plate recognition using YOLOv7 and PaddleOCR. International Advanced Research Journal in Science, Engineering and Technology (IARJSET), Special Issue – ICMART 2023. Retrieved from https://iarjset.com/wpcontent/uploads/2023/06/IARJSET-ICMAR

33 Comparative Design and Analysis of Micro-strip Patch Antenna Geometries for Non-Invasive Sweat Based Dielectric Sensing at 2.4 GhzNor Yusuhanna Mohd Yusoff1,*, Ahmad Rashidy Razali, Sarizan Saaidon 1Department of Electronic & Computer Engineering Technology, Computer Engineering Technology Division, Japan-Malaysia Technical Institute (JMTi), 59, Lorong Perindustrian Bukit Minyak 15, Taman Perindustrian Bukit Minyak, 14100 Simpang Ampat, Penang, Malaysia *Corresponding Author Email: [email protected] 1. Introduction The growing demand for wearable and non-invasive biomedical monitoring systems has driven the development of compact, low-cost, and passive sensing technologies. Antenna-based dielectric sensing has gained increasing attention due to its ability to detect changes in the surrounding medium through variations in electromagnetic characteristics such as resonance frequency and reflection coefficient [1]. This sensing approach is particularly attractive for wearable applications because it eliminates the need for additional sensing elements and allows seamless integration with wireless communication systems [2]. Sweat analysis has emerged as a promising non-invasive method for monitoring physiological conditions related to hydration status and electrolyte balance. Since sweat is primarily composed of water and dissolved salts, variations in its composition result in measurable changes in dielectric properties [3]. Micro-strip patch antennas operating near the skin can exploit these changes by monitoring resonancefrequency shifts. The 2.4 GHz Industrial, Scientific, and Medical (ISM) band is well suited for such applications due to its unlicensed operation, compatibility with common wireless standards, and support for compact antenna designs suitable for wearable use. In addition to sensing performance, sustainability and cost are important considerations in wearable sensor design. Paper-based substrates offer a low-cost, flexible, and environmentally friendly alternative to conventional dielectric materials. With a relatively low dielectric constant, paper substrates support effective antenna operation while enabling disposable or short-term wearable sensing platforms [4]. These characteristics make paper particularly attractive for sustainable antenna-based biomedical sensing applications. Antenna geometry plays a critical role in determining dielectric sensing sensitivity by influencing surface current distribution and fringing-field interaction with nearby materials [5]. Although micro-strip patch antennas have been widely investigated for dielectric and biomedical sensing, direct comparative studies examining the influence of antenna geometry under identical conditions remain limited, especially for paper-based substrates at 2.4 GHz. To address this gap, this study presents a comparative simulation-based analysis of square and circular micro-strip patch antennas for noninvasive sweat-based dielectric sensing. Abstract: Non-invasive and sustainable sensing technologies are increasingly required for wearable biomedical applications, particularly for monitoring physiological conditions through sweat analysis. Antenna-based dielectric sensing has emerged as a promising approach due to its passive operation, low cost, and compatibility with wireless systems. This study presents a comparative simulation-based investigation of square and circular micro-strip patch antenna geometries operating at 2.4 GHz for non-invasive sweat-based dielectric sensing using a paper substrate. The antennas were designed and analyzed using CST Microwave Studio under identical conditions to ensure a fair comparison. Paper was selected as the substrate material owing to its low dielectric constant, flexibility, cost-effectiveness, and environmental sustainability, making it suitable for disposable and wearable sensing platforms. The sensing performance was evaluated under three dielectric loading conditions representing simplified sweat environments: dry (air), plain water, and saline solution. Resonance-frequency shift in the reflection coefficient (S₁₁) response was employed as the primary sensing indicator. Simulation results demonstrate that both antenna geometries are capable of detecting dielectric variations through measurable resonance-frequency shifts. The square micro-strip patch antenna consistently exhibited larger frequency shifts under dry-to-water conditions, indicating higher sensitivity to moisture-related dielectric changes, while both geometries showed comparable responses under saline loading due to the dominant influence of ionic conductivity. The findings provide useful design guidance for sustainable, paper-based antenna sensors in wearable healthcare monitoring applications.Keywords: Micro-strip patch antenna, Dielectric sensing, Sweat-based sensing, Paper-based substrate, Resonance frequency shift

34 2. Literature review Antenna-based dielectric sensing has emerged as a passive and non-invasive technique for detecting changes in the surrounding environment through variations in resonance frequency and reflection coefficient. Micro-strip patch antennas are widely used for this purpose due to their planar structure, low fabrication cost, and compatibility with wireless and wearable systems [6]. Previous studies have demonstrated that dielectric loading near the antenna surface alters the effective electrical length, producing measurable resonancefrequency shifts suitable for liquid, moisture, and biomedical sensing applications [7]. In the context of wearable healthcare, antenna-based sensors have been investigated for sweat and hydration monitoring using simplified liquid models such as water and saline solutions [8]. These studies report that conductive liquids, particularly saline, induce stronger resonance perturbations due to the combined effects of dielectric permittivity and ionic conductivity. At the same time, growing interest in sustainable sensing platforms has led to the use of paper-based substrates, which offer low cost, flexibility, and environmental benefits while supporting effective antenna operation at microwave frequencies [9]. Antenna geometry significantly influences dielectric sensing sensitivity by shaping surface current distribution and near-field interaction [10]. Square patch antennas generally produce stronger edge fringing fields, whereas circular patches exhibit more symmetrical current distributions and stable resonance behaviour. Although these effects have been reported, direct comparative studies evaluating square and circular micro-strip patch antennas on paper-based substrates under identical operating conditions at 2.4 GHz remain limited. This motivates the present work, which focuses on a controlled geometry-based sensitivity comparison for noninvasive sweat-based dielectric sensing. 3. MethodologyThe overall research procedure was structured into four sequential phases: antenna design and parameter calculation, simulation setup and optimization, dielectric loading and verification, and data analysis. This phased approach ensures a systematic and repeatable investigation of antenna sensing behaviour under different dielectric conditions 3.1 Antenna Design and Parameter Calculation In the first phase, square and circular micro-strip patch antennas were analytically designed to operate at the target resonance frequency of 2.4 GHz using a paper-based substrate. Initial antenna dimensions were calculated using classical transmission-line model equations to provide accurate starting parameters prior to full-wave simulation [11]. The width of the micro-strip patch (W) was calculated as: ? = ௖ଶ௙ೝ ටଶఌೝାଵ (1) The effective dielectric constant (????) is then calculated to account for fringing fields at the patch edges and the effective patch length (????) is obtained: ?௘௙௙ =ఌೝାଵଶ +ఌೝିଵଶ(1 + 12 ௛ௐ)ି଴.ହ (2) ?௘௙௙ =௖ ଶ௙ೝඥఌ೐೑೑ (3) Finally, the actual patch length (L) was calculated by subtracting the fringing-field extension (ΔL) from the effective length: ? = ?௘௙௙ − 2Δ? (4) The calculated dimensions served as the initial design parameters for both square and circular antenna geometries. 3.2 Antenna Design and Parameter In phase 2, both antenna geometries consist of a copper radiating patch printed on a paper substrate with a full ground plane beneath it. A 50-Ω micro-strip feed line was used to excite the antenna due to its simplicity and suitability for planar antenna structures [4]. The same substrate material, substrate thickness, ground plane dimensions, and feeding configuration were applied to both antennas to ensure a fair comparison. The antenna geometries are shown in Figure 1. Fig. 1 - Micro-strip patch antenna structure in CST 3.3 Simulation Setup and Optimisation The antenna models were implemented and simulated in CST Microwave Studio using full-wave electromagnetic analysis in phase 3. The simulation frequency range was set from 2.0 GHz to 3.0 GHz to capture resonance behaviour around the 2.4 GHz ISM band. Adaptive mesh refinement was applied, particularly near the patch edges, to improve simulation accuracy [4]. Minor dimensional adjustments were performed when necessary to ensure that both antenna geometries resonated close to 2.4 GHz under dry conditions. This optimisation step ensures that subsequent resonance-frequency shifts are primarily caused by dielectric loading rather than design inaccuracies. 3.4 Dielectric loading conditionsTo emulate sweat-based sensing environments, three dielectric loading conditions were introduced near the antenna surface: dry (air), plain water, and saline solution. These conditions represent simplified models of sweat-related environments commonly adopted in antenna-based dielectric sensing studies [11]. The dielectric properties used in the simulations are summarised in Table 1.

35 Table 1 - Dielectric Properties 3.5 Sensitivity evaluationDielectric sensing performance was evaluated by analysing the reflection coefficient (S₁₁) response of each antenna geometry. The resonance frequency was identified at the minimum S₁₁ point. Sensitivity was quantified using the resonance-frequency shift Δf calculated as: ∆? = ???? − ??????? (5) where ???? is the resonance frequency under dry conditions and ??????? corresponds to the resonance frequency under water or saline loading. Larger resonance-frequency shifts indicate stronger electromagnetic coupling between the antenna near-field region and the surrounding dielectric medium, and therefore higher dielectric sensing sensitivity. 4. Result and discussion4.1 Sensitivity Analysis under Dry and Water ConditionsThe dielectric sensitivity under dry-to-water transition was evaluated using resonance-frequency shift as the primary metric. As shown in Figure 2 and Figure 3, both antenna geometries exhibit noticeable resonance-frequency shifts when water is introduced. The square patch antenna produces a slightly larger shift than the circular patch antenna, indicating stronger electromagnetic coupling with the surrounding dielectric medium [4]. Fig. 2 - Simulated S₁₁ response of the square micro-strip patch antenna under dry and water conditions Fig. 3 - Simulated S₁₁ response of the circular micro-strip patch antenna under dry and water conditions This enhanced sensitivity is attributed to the square geometry, which generates stronger fringing fields along its edges and corners, thereby increasing interaction with nearby dielectric materials. In contrast, the circular patch antenna exhibits a more symmetrical current distribution, resulting in smoother near-field patterns and marginally lower sensitivity [12]. Quantitative resonance-frequency shifts for both antenna geometries under dry and water conditions are presented in Table 2. As shown in Table 2, both antenna geometries exhibit a downward shift in resonance frequency when transitioning from dry to water conditions. The square patch antenna shows a frequency shift of 0.0909 GHz, which is slightly larger than the 0.0860 GHz shift observed for the circular patch antenna, confirming higher moisture sensitivity for the square geometry under identical conditions. Table 2 - Dry vs Water Resonance-Frequency Shift 4.2 Sensitivity Analysis under Dry and Saline Conditions Sensitivity under dry-to-saline transition was analysed to represent electrolyte-rich sweat environments. Figure 4 and Figure 5 show that both antenna geometries exhibit substantial resonance-frequency shifts under saline loading, confirming strong sensitivity to conductive dielectric changes. Fig. 4 - Simulated S₁₁ response of the square micro-strip patch antenna under dry and saline conditions Fig. 5 - Simulated S₁₁ response of the circular micro-strip patch antenna under dry and saline conditions Condition Relative Permittivity (εr)Dry 2.31Plain Water 2.50Saline 2.10Antenna Geometry Dry Resonance (GHz)Water Resonance (GHz)Δf (GHz)Square Patch 2.3988 2.3079 0.0909 Circular Patch2.3951 2.3091 0.0860

36 As summarized in Table 3, both antenna geometries exhibit comparable resonance-frequency shifts under saline loading, with the square patch antenna showing a shift of approximately 0.1089 GHz and the circular patch antenna exhibiting a similar shift of 0.1092 GHz. The minimal difference between these values indicates that under saline conditions, the sensing response is primarily governed by ionic conductivity rather than antenna geometry. The presence of conductive ions intensifies near-field electromagnetic interaction, resulting in strong resonance perturbation for both geometries [13]. Nevertheless, the square patch antenna maintains a sharper and more well-defined resonance minimum in the S₁₁ response, which is advantageous for stable and reliable frequency extraction in sensing applications. Table 3 - Dry vs Saline Resonance-Frequency Shift Antenna GeometryDry Resonance (GHz)Saline Resonance (GHz)Δf (GHz)Square Patch2.3989 2.5078 0.1089Circular Patch2.3951 2.5043 0.10925. Conclusion and recommendationThis study presented a comparative simulation-based evaluation of square and circular micro-strip patch antenna geometries for non-invasive sweat-based dielectric sensing at 2.4 GHz using a paper-based substrate. A controlled framework was employed to isolate the effect of antenna geometry, with sensing performance assessed through resonance-frequency shifts in the reflection coefficient (S₁₁) response under different dielectric loading conditions. Simulation results confirm that both antenna geometries function effectively as passive dielectric sensors, exhibiting clear resonance-frequency shifts under dry, water, and saline environments. The square patch antenna consistently demonstrated higher sensitivity under dry-to-water transition due to enhanced fringing-field interaction, while both geometries showed comparable responses under saline loading, where ionic conductivity dominates the sensing mechanism. Based on the observed sensitivity trends, the square micro-strip patch antenna is identified as the more suitable geometry for sweat-based dielectric sensing, particularly for early detection of moisture-related changes. Future work should focus on experimental validation and assessment under realistic conditions, including real sweat samples, humanbody coupling, and mechanical deformation, to further support the development of wearable antenna-based sensing systems. Acknowledgement The authors would like to express their sincere appreciation to JMTI for the support and guidance provided throughout the course of this research. References [1] Z. U. Islam, A. Bermak, and B. Wang, “A Review of Microstrip Patch Antenna-Based Passive Sensors,” Oct. 01, 2024, Multidisciplinary Digital Publishing Institute (MDPI). doi: 10.3390/s24196355. [2] A. Ashyap, R. Raad, F. Tubbal, W. A. Khan, and S. Abulgasem, “Comprehensive Review of Wearable Antennas With Flexible Periodic Structures for Body-Effect Mitigation,” IEEE Access, vol. 13, no. January, pp. 22590–22636, 2025, doi: 10.1109/ACCESS.2025.3536525. [3] S. H. Rakib, M. T. Reza, and M. F. Islam, “Design of microstrip patch sensor for non-invasive body electrolyte monitoring,” 2020 IEEE Reg. 10 Symp. TENSYMP 2020, no. June, pp. 1110–1113, 2020, doi: 10.1109/TENSYMP50017.2020.9230891. [4] A. M. Mozi et al., “Sensitivity Study of Slotted Radiator Antenna-Based Sensor for Sweat Monitoring,” Proc. - 2022 RFM IEEE Int. RF Microw. Conf. RFM 2022, vol. 2, no. 1, pp. 1–4, 2022, doi: 10.1109/RFM56185.2022.10065148. [5] B. Nataraj and K. R. Prabha, “Analysis of various Microstrip Patch Antenna Designs for 5G Applications,” in 2022 1st International Conference on Electrical, Electronics, Information and Communication Technologies, ICEEICT 2022, Institute of Electrical and Electronics Engineers Inc., 2022. doi: 10.1109/ICEEICT53079.2022.9768642. [6] M. J. Hakeem and M. M. Nahas, “Improving the Performance of a Microstrip Antenna by Adding a Slot into Different Patch Designs,” Eng. Technol. Appl. Sci. Res., vol. 11, no. 4, pp. 7469–7476, 2021, doi: 10.48084/etasr.4280. [7] M. T. Khan, X. Q. Lin, and C. Zhe, “Slotted Microstrip Patch Antenna for Improved Sensing of Moisture Content in Leaf,” 2024 13th Int. Conf. Commun. Circuits Syst., pp. 250–255, 2024, doi: 10.1109/icccas62034.2024.10652684. [8] J. Karthi, S. Venu Aravind, N. Sreejith, and M. Palanivelan, “Design and Development of an EcoFriendly Antenna Using a Paper-Based Substrate for 5G IoT Devices,” Proc. - 3rd Int. Conf. Artif. Intell. Mach. Learn. Appl. Healthc. Internet Things, AIMLA 2025, pp. 1–5, 2025, doi: 10.1109/AIMLA63829.2025.11040480. [9] M. Ahmad et al., “Paper-Based Printed Antenna: Investigation of Process-Induced and ClimaticInduced Performance Variability,” Adv. Eng. Mater., vol. 25, no. 16, pp. 1–10, 2023, doi: 10.1002/adem.202201703. [10] J. Rodrigues, J. C. Sá, F. J. G. Silva, L. P. Ferreira, G. Jimenez, and G. Santos, “A rapid improvement process through ‘quick-win’ lean tools: A case study,” Systems, vol. 8, no. 4, pp. 1–19, 2020, doi: 10.3390/systems8040055.

37 [11] A. A. Bakar, A. Ibrahim, A. M. Mozi, N. M. Faudzi, A. R. Razali, and N. H. A. Bakar, “Slotted Antenna Based Sensor for Liquid Sensing,” in 2024 IEEE Asia-Pacific Conference on Applied Electromagnetics, APACE 2024, Institute of Electrical and Electronics Engineers Inc., 2024, pp. 413–416. doi: 10.1109/APACE62360.2024.10877340. [12] A. M. Mozi, A. R. Razali, N. H. A. Rahman, S. N. Azemi, N. M. Faudzi, and A. Ibrahim, “Sensitivity Study of a Coplanar Waveguide Antenna-Based Sensor for Hydration Monitoring,” in 2024 IEEE 7th International Conference on Electrical, Electronics and System Engineering (ICEESE), IEEE, Nov. 2024, pp. 1–4. doi: 10.1109/ICEESE62315.2024.10828554. [13] H. I. Hamd, W. Q. Mohamed, and H. B. Ahmed, “Design and Simulation of H Shape and Duplicate U Shape Slots Microstrip Patch Antenna for WiMAX Applications,” in ISMSIT 2021 - 5th International Symposium on Multidisciplinary Studies and Innovative Technologies, Proceedings, Institute of Electrical and Electronics Engineers Inc., 2021, pp. 278–281. doi: 10.1109/ISMSIT52890.2021.9604637.

38 A study on combustion characteristics of water-emulsified with new generation bio diesel fuelRyo YAMAIZUMI (†: presenter) Department of Maritime Technology, Hiroshima College, National Institute of Technology, Japan [email protected]: Internal combustion engine, hydrotreated vegetable oil, water-emulsified fuel Abstracts: Recently, reducing carbon dioxide emissions has become an urgent global issue due to the accelerating impacts of climate change. In the field of internal combustion (IC) engines, alternative energy sources such as biofuels, hydrogen, and ammonia, as well as electrification, have been studied across various industries, including the maritime sector, to achieve a decarbonized society. However, the use of hydrogen and ammonia requires control of abnormal combustion and new infrastructure, while electrification faces challenges related to CO₂ emissions from battery production and recycling. Therefore, this study focuses on hydrotreated vegetable oil (HVO), a renewable fuel derived from waste cooking oil, with excellent ignitability and low aromatic content. Although HVO can be used in diesel engines without modification, its NOₓ emissions are comparable to those of conventional diesel fuel. Also, water-emulsified fuels are known to suppress NOₓ emissions through the latent heat of vaporization; however, diesel–water emulsions often suffer from poor ignition stability. In this study, the effectiveness of HVO–water emulsion fuel is investigated to achieve a balance between high thermal efficiency and reduced NOₓ emissions. Steady-state experiments at constant engine speed and load are conducted to investigate the effects of water content on engine performance and exhaust emissions. † Fig.1 The concept of micro-explosion phenomenon

39 Analysis of Warm Up and Cold Down Processes in Marine Engine Operation TrainingKazumasu Takigawa (†: presenter) Department of Maritime Technology, Hiroshima College, National Institute of Technology, Japan [email protected]: Marine Engine, Cold down, Warm up Abstracts:This presentation focuses on the analysis of warmup and cool-down processes during marine engine operation training conducted at KOSEN. The purpose of this study is to clarify the engineering and educational importance of these processes, which are essential for ensuring safety, efficiency, and mechanical stability in marine engines. During the warm-up phase, gradual temperature increase allows for proper lubrication, thermal expansion, and stabilization of engine components before full operation. Conversely, the cool-down phase prevents thermal shock and metal fatigue by allowing the temperature to decrease slowly, ensuring long-term engine reliability.† Fig.1 Starting and Turning off Diesel Fig.2 Simulation of cold/warm-start transients

40 Honda's Engineering Marvels: The Pride of JapanHiroto Omasa (†: presenter) Department of Distribution and Information Engineering, Hiroshima College, National Institute of Technology, Japan [email protected]: Honda Motor Co., VTEC, ASIMO, Honda Jet, Innovation, Power of DreamsAbstracts: Since its founding in 1948, Honda Motor Co., Ltd. (Honda) has become one of Japan's most esteemed companies, creating innovative products driven by a strong belief in \"using technology to help people\". The driving force behind this is the spirit of dreams and challenges, symbolized by their slogan, \"The Power of Dreams\". This presentation will explain Honda's innovation by highlighting three iconic examples from its diverse technological fields. First is \"VTEC\", a cornerstone of automotive technology. This revolutionary mechanism achieves two conflicting goals—fuel efficiency and usability at low RPM, and overwhelming power at high RPM—by switching valve timing and lift according to engine speed. Second is their robotics technology, represented by the humanoid robot \"ASIMO\". Its realization of bipedal walking and autonomous behavior demonstrated to the world the potential of robots as partners coexisting in human society. Third is their entry into the aviation business with the \"Honda Jet\". By adopting a unique Over-TheWing Engine Mount configuration, it achieved bestin-class speed and fuel efficiency, revolutionizing the very light jet market. These relentless challenges in the respective fields of \"Land\", \"Human\", and \"Air\" are prime examples of how Honda has turned \"dreams\" into reality through \"technology\". Through this presentation, I aim to convey the spirit of Japanese monozukuri (manufacturing) and the high level of its technological prowess. †

41 Graduation Research in Technical College Education using OpenFOAMKangbin LEI (†: presenter) Department of Maritime Technology, Hiroshima College, National Institute of Technology, Japan [email protected]: OpenFOAM, Technical College Education, Fluid MechanicsAbstracts: Fluid mechanics class is a specialized basic subject in engineering education that is difficult to learn for technical college students and to earn the class credits, the study of students on fluid mechanics is easily limited to taking only the form of partial differential equations into account, and there is a problem in the quality of education that students are not able to understand and apply their knowledge of fluid mechanics very well. The purpose of this study is to introduce OpenFOAM that is a free software for fluid analysis into the classroom, which has been attracting attention in the field of education, in response to the question of quality assurance in fluid mechanics classes in technical college education, and to find out how the simulation results of fluid phenomena visualized by students themselves can be connected to theoretical knowledge of fluid mechanics. From this educational practice in technical college, we aim to develop a learning method to draw students' interests and make them feel \"interested and want to know more\" by introducing the OpenFOAM into fluid mechanics’ class. In this paper, we introduce a practical case study of a fluid simulation theme using OpenFOAM at Hiroshima College, in which students were asked to simulate fluid dynamics on their own by trial and error, following the operation manual of OpenFOAM, even though they know less about theoretical knowledge of fluid mechanics. † Fig.1 3D model and mesh generation of carFig.2 Streamlines of flow around the car

42 CULTURAL STUDIES CATEGORY