Design Of Integration Model Of SPC/EPC For Complex Products Based On Bayesian Methods

Manufacturing of Complex products and related technology is important bases for national economy and security, and is also important symbol of comprehensive national power such as national industrial foundation, economic and technological level, and so on. One important problem complex products manufacturing industry faces with is how to improve quality management and control level. Production type of most complex products is multi-variety and small-batch, and their production process is non-linear and dynamic, which brings difficulties to process control.Statistical Process Control and Engineering Process Control are two different methods for process control. They are playing important roles and they have complementary advantages. This article designed an integration model of SPC/EPC according to the non-linear and dynamic properties of complex products. As for EPC model, a process adjustment method based on Bayesian state space model was chosen, according to the properties of complex products manufacturing process that disturbance series is difficult to fit. As for SPC model, first, a model suitable for double boundary process based on the Bayesian algorithm was designed, including Bayesian estimation of process parameters, the selection of threshold and abnormality-judging standards, And then the properties and influencing factors of Bayesian estimation was figured out. And then it figured out the properties and influencing factors of Bayesian estimation. This paper pointed out that in statistical process control,the main factor affecting the accuracy of Bayesian algorithm was the ratio of state variance to observed variance,but not prior variance. The control scheme was tested by Matlab simulation using ARL (Average Run Length) and empirical study. The applicable condition of Bayesian estimation in statistical process control was also given. Finally, Shewhart control chart was also added into SPC model as a supplement to detect big fluctuation, and the integrated model was summarized in detail and the running rules was designed. This paper provided new solutions for process control of complex products.