VIETNAM NATIONAL UNIVERSITY HO CHI MINH CITY HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY NGUYEN VAN TRIEU VY APPLICATION OF ARTIFICIAL INTELLIGENCE ALGORITHMS FOR PREDICTING OVERALL EQUIPMENT EFFECTIVENESS AT AN ASSEMBLY MANUFACTURING FACTORY Major: Industrial Engineering Major code: 8.17 MASTER’S THESIS HO CHI MINH CITY, June 2024 THIS THESIS IS COMPLETED AT HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY – VNU-HCM Supervisor: Dr. Nguyen Duc Duy Examiner 1: Assoc. Phan Thi Mai Ha Examiner 2: Dr. Do Thanh Luu This master’s thesis is defended at HCM City University of Technology, VNU- HCM City on June 08th, 2024 Master’s Thesis Committee: 1.
Do Ngoc Hien 2. Dinh Ba Hung Anh 3. Phan Thi Mai Ha 4. Do Thanh Luu 5.
Nguyen Van Thanh Approval of the Chair of Master’s Thesis Committee and Dean of Faculty of Mechanical Engineering after the thesis being corrected (If any). CHAIR OF THESIS COMMITTEE DEAN OF FACULTY OF MECHANICAL ENGINEERING VIETNAM NATIONAL UNIVERSITY - HO CHI MINH CITY SOCIALIST REPUBLIC OF VIETNAM HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY Independence – Freedom - Happiness THE TASK SHEET OF MASTER’S THESIS Full name: Nguyen Van Trieu Vy Student ID: 2170634 Date of birth: Nov 4th, 1999 Place of birth: Dong Nai Province Major: Industrial Engineering Major ID: 8520117 I. THESIS TITLE (In Vietnames): ÁP DỤNG CÁC THUẬT TOÁN TRÍ TUỆ NHÂN TẠO TRONG DỰ ĐOÁN HIỆU SUẤT THIẾT BỊ TỔNG THỂ TẠI NHÀ MÁY SẢN XUẤT LẮP RÁP II. THESIS TITLE (In English): APPLICATION OF ARTIFICIAL INTELLIGENCE ALGORITHMS FOR PREDICTING OVERALL EQUIPMENT EFFECTIVENESS AT AN ASSEMBLY MANUFACTURING FACTORY III.
TASKS AND CONTENTS: - To study about the artificial intelligence algorithms and application of them to manufacturing field. - To deeply understand the data flow in manufacturing company where the thesis is applied. - To propose the mathematical models produced by machine learning to apply for the case study and validate the effectiveness. THESIS START DAY: V.
THESIS COMPLETION DAY: June 08th, 2024 VI. Nguyen Duc Duy Ho Chi Minh City, June 08th, 2024 SUPERVISOR HEAD OF DEPARTMENT (Full name and signature) (Full name and signature) DEAN OF FACULTY OF MECHANICAL ENGINEERING (Full name and signature) i ADKNOWLEDGEMEMTS As I complete this thesis, I wish to acknowledge the divine presence that guided me throughout this scholarly endeavor. My sincerest gratitude to the Almighty for granting me the strength, wisdom, and determination to bring this work to fruition. I acknowledge the blessings that have enabled me to pursue knowledge and produce this thesis.
I would like to express my gratitude to my supervisor, Dr. Nguyen Duc Duy, whose guidance, patience, and expertise were invaluable throughout this research journey. Thank you for your unwavering support and insightful feedback. I would like to thank my parents, sibling, and extended family, thank you for your unwavering encouragement and understanding during late nights and weekends spent writing.
Your love and belief in me kept me going. I extend my deepest appreciation to the friends who stood by me during this academic odyssey. Their unwavering support, late-night brainstorming sessions, and shared laughter made the journey memorable. I am grateful for your friendship and encouragement.
You turned the solitary pursuit of knowledge into a collaborative adventure. To all my friends, near and far, thank you for being my pillars of strength. This thesis bears witness to our collective resilience and camaraderie. In closing, this thesis is a collective effort—a tapestry woven with threads of guidance, collaboration, and resilience.
I am honored to have walked this path, and I look forward to contributing further to the field of industrial system engineering. With heartfelt gratitude, Nguyen Van Trieu Vy June 2023 ii ABSTRACT The Overall Equipment Effectiveness (OEE) stands as a key performance metric widely adopted in the manufacturing industry, aiding in enhancing productivity. This metric offers a comprehensive overview to higher man-agement, enabling them to identify equipment-related losses. With the ad-vancements in Industry 4.0 technologies, the Manufacturing Execution Sys-tem facilitates real-time data collection, enhancing production efficiency and reducing manufacturing costs.
However, the real-time depiction of OEE of-ten fails to provide decision-makers with timely insights to carry out their tasks effectively. This study formulated machine learning models to tackle this challenge by forecasting the OEE for the upcoming working shift. First-ly, the historical dataset encompasses 31 features collected and processed to estimate the OEE value. Then, prominent machine learning models were uti-lised as prediction models: Linear Regression, Support Vector Regression, Random Forest, Extreme Gradient Boosting, and Artificial Neural Networks.
The results show that the Extreme Gradient Boosting performs well for the OEE prediction with accuracy in training higher than 99% and testing nearly 90%. Our study illustrates an actionable knowledge-discovery process using a real-world data mining approach for the manufacturing industry, potential-ly applicable to other sectors. iii TÓM TẮT LUẬN VĂN THẠC SĨ Hiệu quả thiết bị tổng thể (OEE) là một chỉ số hiệu suất quan trọng được áp dụng rộng rãi trong ngành sản xuất, giúp nâng cao năng suất. Chỉ số này cung cấp một cái nhìn tổng quan cho quản lý cấp cao, cho phép họ xác định các tổn thất liên quan đến thiết bị.
Với sự phát triển của các công nghệ Công nghiệp 4.0, Hệ thống Thực thi Sản xuất (MES) hỗ trợ thu thập dữ liệu theo thời gian thực, nâng cao hiệu quả sản xuất và giảm chi phí sản xuất. Tuy nhiên, việc hiển thị OEE theo thời gian thực thường không cung cấp cho người ra quyết định những thông tin kịp thời để thực hiện nhiệm vụ của họ một cách hiệu quả. Nghiên cứu này đã xây dựng các mô hình học máy để giải quyết thách thức này bằng cách dự báo OEE cho ca làm việc sắp tới. Đầu tiên, tập dữ liệu lịch sử bao gồm 31 đặc điểm được thu thập và xử lý để ước tính giá trị OEE.
Sau đó, các mô hình học máy nổi bật đã được sử dụng làm mô hình dự đoán: Linear Regression, Support Vector Regression, Random Forest, Extreme Gradient Boosting, and Artificial Neural Networks. Kết quả cho thấy Extreme Gradient Boosting hoạt động tốt cho dự báo OEE với độ chính xác trong huấn luyện cao hơn 99% và kiểm tra gần 90%. Nghiên cứu của chúng tôi minh họa một quy trình khám phá tri thức có thể hành động bằng cách sử dụng phương pháp khai thác dữ liệu thực tế cho ngành sản xuất, có thể áp dụng cho các lĩnh vực khác. iv THE COMMITMENT OF THE THESIS’S AUTHOR I ensure that my thesis, “APPLICATION OF ARTIFICIAL INTELLIGENCE ALGORITHMS FOR PREDICTING OVERALL EQUIPMENT EFFECTIVENESS AT AN ASSEMBLY MANUFACTURING FACTORY”, is conducted directly throughout the entire research process.
All data used for the topic has a clear origin and has not been published elsewhere. If there is any discrepancy in the data used, I take full responsibility. Nguyen Van Trieu Vy v CONTENTS THE TASK SHEET OF MASTER’S THESIS. iii TÓM TẮT LUẬN VĂN THẠC SĨ.
vi TABLE OF FIGURES. viii TABLE OF TABLES .1 Overall Equipment Effectiveness Metric .2 OEE prediction research .1 Overall Equipment Effectiveness .2 AI-based OEE prediction models. Support Vector Regression [26]. Random Forest Regression [26].
Extermme Gradient Boosting Regression [26]. Artificial Neural Network [26] .4 Model Performance Metrics .4 Modeling and evaluation. Support Vector Regression. Random Forest Regression .6 Comparison performance of models.
CONCLUSION AND FUTURE DIRECTION .1 Conclusion and future direction. 57 vii TABLE OF FIGURES Figure 1. Production dashboard using in daily meeting with OEE visulization on the top-right corner. OEE measurement tools and industry integration aspects [1].
Step by step to process data. Prediction approach of the thesis. Support vector Machine model in 2-dimensional space. Role of Kernel function in SVM/SVR.
Random Forest Regression model.XGBoost Regression model. Feedforward in ANN. Backpropagation in ANN. Cross-industry Standard Process for Data mining.
S-line production layout. Data process flow of the case study. Data processing to create label dataset. Example of related-time data.
Example of related-time data after processing. Features dataset structure. Data process flow to create feature dataset. Data modeling flow.
Prediction result of Linear Regression model. Prediction result of SVR model using “linear” kernel. Prediction result of SVR model using “polynominal” kernel. Prediction result of SVR model using “sigmoid” kernel.
Prediction result of SVR model using “rbf” kernel. Prediction result of Random Forest Regression model. Prediction result of XGBoost Regression model. Prediction result of ANN model.
Mean absolute error comparision among prediction models. Root mean squared error comparision among prediction models. Mean squared error comparision among prediction models. Mean absolute percentage error comparision among prediction models.52 ix TABLE OF TABLES Table 2-1.
Overview of OEE prediction literatures. Column list of “Cycle time” data field. Column list of “Downtime” data field. Column list of “Line error” data field.
Data collected before modeling. Linear Regression evaluation. Random Forest evaluation. XGB Regression evaluation.1 Rationale Nowadays, many businesses focus on digital technology development, such as adopting the Internet of Things (IoT), real-time data access and Cyber Physical Systems [1].
Especially in manufacturing, Industry 4.0 offers a more intercon-nected and comprehensive approach that empowers managers to control and understand aspects of operations better, increasing productivity, improving processes and accelerating growth [2-4]. Moreover, increasing customer demand satisfaction is necessary to achieve better revenue while minimising costs: investment, quality cost, labour cost, and maintenance costs are very important in a global market. As a result, decisions must always be made promptly and effectively through the production process in which equipment is a critical factor. Overall Equipment Effectiveness (OEE), developed based on the concept of Total Productive Maintenance (TPM) proposed by Nakajima in 1988, is a critical metric in manufacturing.
It measures overall equipment performance and highlights areas for improvement and optimization, emphasizing continuous enhancement of component parameters until optimal OEE is achieved. TPM aims to minimize machine downtime and reduce defects primarily caused by equipment. As a result, it enhances overall production line efficiency, reduces costs, minimizes inventory, and simultaneously improves labor productivity [5]. The primary benefit of using the OEE metric in manufacturing is to enhance return on investment (ROI) [5].
Machinery represents a significant investment for companies, and they continually seek strategies to maximize their ROI. By measuring OEE data, businesses can demonstrate the value of investing in their machinery and equipment systems. Initially, during the early production stages, equipment utilization efficiency contributes only a small portion to a company’s overall profits. 1 However, as production scales up and investments in machinery increase, minimizing waste and losses leads to higher overall profitability.
As production processes evolve, increased scale and additional investments in machinery result in reduced waste and losses. Consequently, businesses reap greater profits. OEE also aids in increasing competitiveness within the industry. Efficiently managed equipment translates to improved competitiveness.
By addressing inefficiencies, companies can optimize their asset utilization and reduce the need for additional capital investment. In summary, OEE serves as a compass for continuous improvement, allowing businesses to make informed decisions and enhance their overall performance. OEE also helps businesses enhance competitiveness within the same industry. For production operations, reducing waste translates to improved competitive capabilities.
If a production line lacks efficiency, companies need the necessary processes and methods to maximize their infrastructure. Based on the information obtained through the OEE metric, managers and analysts can identify any constraints or bottlenecks in production. Additionally, the OEE metric helps production managers visually represent the performance of the production line. OEE allows businesses to easily grasp the production efficiency.
This is achieved by using calculated formulas and observing any production losses. Each production factor, such as availability, performance, and quality, is clearly expressed through detailed numerical values. It enables companies to assess the current production status and identify areas that require timely improvement. Finally, senior managers can gain detailed production insights to devise timely measures for factory fluctuations.
Businesses cannot improve production operations by merely stumbling around.
