| Summary: | 碩士 === 輔仁大學 === 統計資訊學系應用統計碩士班 === 105 === Because control chart patterns (CCPs) are typically associated with specific root causes that antagonistically upset the process, the effective recognition of CCPs is crucial for process improvement. Accordingly, the research issue of recognition of CCPs has drawn considerable attention in recent years. In addition, because the use of engineering process control (EPC) can greatly improve the statistical process control (SPC) process, the success of integration of an SPC-EPC or a multiple inputs multiple outputs (MIMO) systems has been widely reported. However, even though numerous studies have addressed an increased use of SPC-EPC mechanism, there has been very little research discussed on the recognition of CCPs for a MIMO system. It is much more difficult to recognize the CCPs in a MIMO system since two or more disturbances are simultaneously involved in the process. The purpose of this study is thus to propose several machine learning (ML) techniques to overcome the difficulties for recognition of embedded CCPs in a MIMO system. Because of their efficient and fast algorithms and effective classification performance, the proposed ML classifier include artificial neural networks (ANN), support vector machine (SVM), extreme learning machine (ELM), and multivariate adaptive regression splines (MARS). Additionally, in contrast to using the typical process outputs alone in a classifier, this study employs both process outputs and EPC compensation to assure the effectiveness of CCPs recognition. Experimental results revealed that the proposed approaches are able to effectively recognize various CCPs for a MIMO system.
|
|---|
