Date of Award
8-2020
Degree Name
Doctor of Philosophy
Department
Industrial and Entrepreneurial Engineering and Engineering Management
First Advisor
Dr. Lee Wells
Second Advisor
Dr. Azim Houshyar
Third Advisor
Dr. James Burns
Fourth Advisor
Dr. Ikhlas Abdel-Qader
Keywords
Statistical process control, image simulation, control chart, circulant embedding, MPCA, spatial cross correlation
Abstract
The use of image data in manufacturing quality control systems has grown rapidly over the years. Technological advancements in imaging devices (e.g., digital cameras, x-ray scanners, infrared cameras) have provided convenient and cost-effective ways to obtain images. These images contain important quality features (e.g., textures and product dimensions) that can be used to monitor, diagnose, predict, and control manufacturing processes. A key tool for monitoring images is statistical process control charts. Integrating images of parts to detect process shifts can significantly improve a quality systems efficiency. Unfortunately, appropriate techniques that take full advantage of data-rich images for control charting are not well-developed, and more applications can be investigated by utilizing images in this area.
The overarching goal of this research is to investigate new opportunities of using high-dimensional image data with control charts and to increase their performances in real-world applications. This research addresses the need to extend current control charting approaches and proposes a new framework for using image data for manufacturing quality control. Specifically, a multi-image monitoring framework is proposed to monitor multiple images of the same part simultaneously in order to improve the overall system performance. In this dissertation, two approaches toward implementing the multi-image monitoring framework are introduced and the performance gains obtained with these approaches over monitoring single images are presented. In addition, in order to quantify these performance gains, a computationally efficient approach to simulate cross-correlated images was developed. Both simulation and experimental results justify the effectiveness of the proposed framework.
Access Setting
Dissertation-Open Access
Recommended Citation
Chen, Shengfeng, "A Framework for Using Fused Image Data in Statistical Process Control" (2020). Dissertations. 3652.
https://scholarworks.wmich.edu/dissertations/3652