Application of Machine Learning Approaches to the Recognition of Control Chart Patterns for an SPC-EPC Process

• Main Authors: Wang, Yi-Hsieh, 王怡仙
• Other Authors: Shao, Yueh-Jen
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/90477567557862319006

碩士 === 輔仁大學 === 統計資訊學系應用統計碩士班 === 103 === In recent year, the integration of statistical process control (SPC) and engineering process control (EPC) has widely used in the industrial processes. Although the integration of SPC and EPC has a great benefit to a process, it causes the problem of recognition of control chart patterns (CCP). That is, EPC is able to compensate for the underlying disturbance; however, it may embed the effects of underlying disturbances. As a result, it becomes much more difficult to recognize the CCP. The recognition of CCP is crucial for the process improvement since those patterns are usually associated with some specific assignable causes. Accordingly, the issue of rapid and correct recognition of CCP for a SPC-EPC process is a very promising research topic in the industry. There has been little research conducted on the recognition of CCP for a SPC-EPC process so far. This study is motivated to propose six machine learning approaches to recognize the mixture patterns of process disturbances. Those six approaches include the artificial neural network (ANN), support vector machine (SVM), extreme learning machine (ELM), time-delay neural network (TDNN), rough set (RS) and random forest (RF). Experimental results reveal that the proposed TDNN scheme has the best performance to recognize the CCP for an SPC-EPC process.