IJMO 2024 Vol.14(4): 129-134
DOI: 10.7763/IJMO.2024.V14.861
Integration Model of Residual-Based Mixed CUSUM-EWMA Chart with Deep Learning-Based Automatic Optical Inspection
Luh Juni Asrini1,2,* and Kung-Jeng Wang2
1 Department of Industrial Engineering, Widya Mandala Surabaya Catholic University, Surabaya, Indonesia
2 Department of Industrial Management, Artificial Intelligence for Operations Management Research Center,
National Taiwan University of Science and Technology, Taiwan
Email: juniasrini@ukwms.ac.id (L.J.A.); kjwang@mail.ntust.edu.tw (K.-J.W.)
*Corresponding author
Manuscript received October 16, 2023; revised January 5, 2024; accepted June 25, 2024; published November 17, 2024
Abstract—Challenges arise when it comes to identifying flaws in small-scale electronic components swiftly during quality inspections. While Convolutional Neural Networks (CNNs) are effective at detecting defects in Automatic Optical Inspection (AOI) systems, their primary focus is on individual samples and lacks the ability to provide real-time information about the production process for process control and monitoring. To address this, a combination of CNNs and statistical process control models can be employed to enable proactive quality inspection in high-speed production lines. By combining a control chart with CNNs, the system showcases outstanding detection performance for even slight variations in quality, as evidenced by the average run length falling within a specific range of shifts. This control chart has been effectively implemented in the manufacturing process of electronic conductors, allowing for systematic quality inspection of minute electronic components on a high-speed production line. The integration of the CNN-based AOI model and the residual mixed multivariate Cumulative Sum-Exponentially Weighted Moving Average (CUSUM-EWMA) model control chart enables simultaneous quality assessment at the individual product level and process monitoring at the system level, leading to efficient detection of defects. The novelty of this research lies in the innovative process control framework that merges the CNN-based AOI model with a residual-based mixed multivariate cumulative sum and exponentially weighted moving average control chart. This integration facilitates real-time monitoring of multivariate autocorrelated processes and enables effective identification of defects.
Keywords—autocorrelated process, automatic optical inspection, deep learning, residual control chart
[PDF]
Cite: Luh Juni Asrini and Kung-Jeng Wang, "Integration Model of Residual-Based Mixed CUSUM-EWMA Chart with Deep Learning-Based Automatic Optical Inspection," International Journal of Modeling and Optimization, vol. 14, no. 4, pp. 129-134, 2024.
Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited
(CC BY 4.0).
