| Summary: | 碩士 === 輔仁大學 === 統計資訊學系應用統計碩士班 === 101 === With the development of technology, high quality of products is essential for a company. Statistical process control (SPC) chart is one of the most important monitoring tools in detecting process disturbances. When a SPC chart triggers a signal, it means that the process is under the situation of out of control. The operator has to find the root causes for the process disturbances. Although, multivariate SPC charts have excellent performance in monitoring a multivariate process, the determination for source of process disturbances is much more difficult than the case of a univariate process. The main reason is that the multivariate process has more than two quality characteristics. When the assignable causes have occurred in a multivariate process, we cannot distinguish which quality characteristic has gone wrong. To overcome this difficulty, this study proposes the hybrid methods of LR-ANN, MARS-ANN and ANOVA-ANN. The use and the superior performances of the propose approaches is demonstrated by conducting a series of simulations.
|
|---|
