关键词: 统计过程控制, EWMA控制图, 残差T²控制图

Abstract: Against the background of rare earth separation enterprises, this paper summarizes the feature and quality attribute of co-product first. Then the exponentially weighted moving average(EWMA)control chart of the single product and the multiple residual T² control chart of the co-product are drawn respectively. The comparative analysis results of the two control charts show that the multiple residual T² control chart of co-product can reduce the probability of false alarm compared to EWMA control chart. Finally, based on the abnormal patterns of multiple residual T² control chart, this paper classifies the abnormal patterns using support vector machine model(SVMM), finds out the classification rules, constructs PSO-SVM classifier and optimizes the SVM parameters by the method of Particle Swarm Optimization Algorithm. The results show that the PSO-SVM classifier can greatly improve the classification accuracy. Pattern recognition can be used to diagnose the causes for the abnormality of the co-product multiple residual T² control chart and improve the level of process quality management.

Key words: statistical process control, exponentially weighted moving average control chart, residual T² control chart