Simultaneous Monitoring of Weibull Parameters by Using Support Vector Regression

ABSTRACT In industrial quality control, tracking process variation is essential in maintaining product reliability, particularly when faced with non‐normal data distributions. This research presents a new adaptive Max‐EWMA control chart combined with Support Vector Regression (SVR) for concurrent monitoring of process mean and variance. The new approach utilizes a dynamic smoothing constant forecasted by SVR to increase sensitivity to real‐time process shifts. By converting Weibull‐distributed data into a normal form, this approach efficiently overcomes the difficulties caused by non‐normality. Exhaustive tests with various kernel functions prove the better performance of the linear kernel in achieving sensitivity‐stability balance, which renders it an ideal choice for industrial applications. The results show the flexibility and reliability of the presented control chart to detect small to moderate changes in process parameters.