| Summary: | 碩士 === 國立雲林科技大學 === 工業工程與管理研究所碩士班 === 100 === Support vector regression is a regression technique based on a support vector machine, and on the regression prediction error, and it is small compared to the least squares method so that it has a better predictive ability. In a regression residuals control chart, the general use of traditional regression is into the control chart, but in recent years, some scholars have found that support vector regression in the regression residual control chart is superior to the predictive ability of the traditional regression support vector regression applications. However, a regression residual control chart uses the traditional three times the standard deviation, and the application of information to a linear type. It lacks for non-linear applications in data types, and is ineffective in monitoring small offsets. Therefore, this study will increase the nonlinear types of data of the EWMA control chart and select the combination of support vector regression, and regression control chart monitoring capabilities in the process shift. The simulated data used in this study support vector regression of the EWMA control chart with the traditional return of the EWMA control chart monitoring to make evaluations, respectively, for different offsets, as measured by the average run length. The study results show that support vector regression application of the linear and nonlinear data is, superior to the traditional regression prediction accuracy rate, and the two regression methods EWMA control chart, in order to support vector regression based on the EWMA control chart of the monitoring capabilities compared to the traditional return of the EWMA control chart, is excellent.
|
|---|
