Dynamic Design And Assessment Of Multivariate Quality Control Chart

The extensive application of advanced multi-quality control technologies and methods is of great significance for enterprises to improve the quality of products with muli-quality characteristic and product competitiveness. How to use quality engeneering technologies to design and produce products with low cost, short period, high quality and high reliability has become the issue of the majority of theoretical researchers and practical workers concern. The main stream of modern multivariate quality engineering is to reduce, suppress and control variation in product forming process.Variation is the fundamental cause of poor quality. Reducing and controlling variation has become a core research domain in modern quality engineering science. Multiple quality control chart is used to distinguish normal and abnormal variation, timely warning, reduce defective products and improve efficiency in multi-quality produce process by quality data analysis and charting. However, the traditional static quality control chart could not meet the diversity of enterprise needs, in order to improve the efficiency and effectiveness of control, it is necessary to improve the existing control charts.This paper focuses on the multiple quality control problems of comlex products with multiple quality characteristics in the early stage of smalll-scale production of new products,the initial stable stage of CTQ definition and the stage of continuous growth of mass production, and from the perspective of dynamic design theory of multivariate quality control charts, this paper focuses on the Bayesian dynamic recursive control chart, the identification of critical quality characteristic and the multiple dynamic quality control chart:(1)The multiple dynamic quality control chart in the small batch production situation is designed and researched. By seclecting the multiple linear mathematical model suitable for Bayesian theory, the inverse Gamma distribution and the multivariate normal distribution are chosen as the prior distribution of the covariance and the mean value, then the parameters of the model are estimated by Bayesian method based on determination of the prior distribution parameters, a chi-square statistic is designed and a multivariate Bayesian dynamic recursive control chart with multiple varieties and small batches is established.The influence factors of control chart performance of multivariate Bayesian recurrent control chart is analyzed by KQS covariance control chart and MEWMA control chart, and the approaches and suggestions to improve the Bayesian estimation efficiency are proposed.Finally, the conclusion of the performance evaluation and optimization of multivariate Bayesian control chart is verified by an example of a multi-quality control process of a company's mobile phone speaker produce, which shows the correctness of this paper'sconclusion.(2) After the initial stability of the product quality, the relationship between CTQ and the control performance of the multiple quality control chart is studied. The inverse matrix method of Mahalanobis-Taguchi System (MTS) method, combined with orthogonal table and signal-noise ratio, is used to describe the application of Mahalanobis-Taguchi System(MTS) method and effect variance method in the research of CTQ, in order to better explain, a kind of mobile phone speaker is used as an example in this paper and the CTQ of this product is optimized by Mahalanobis-Taguchi System (MTS) method and effect variance method, making 22 CTQs of this product to be reduced. In this process, it is also noted that CTQs are different using different CTQ identification methods, ARL is used as the control chart performance evaluation index in this paper to discribe the relationship of CTQ and multi-quality control chart performance, and the choice of strategies and recommendations are given in the actual CTQ identification process.(3)The multiple dynamic control charts are designed and the control performance is evaluated in the mass production stage with increasing prodution. A model of the performance analysis on multivariate adaptive control chart is established and a Matlab program of Markov chain approach and Monte Carlo simulation for it are designed. The dynamic Hotelling T2 control, MEWMA control chart and MCUSUM control chart are compared in order to provide corresponding suggestions for the reasonable selection of multiple dynamic control charts under different conditions. By comparing the model output values, it can be seen that the optimal switching strategy of multiple dynamic control charts are to often and will bring practical problems, therefore, the VSI MAEWMA control chart is designed and the performance is analyzed, it is found that the VSI MAEWMA control chart can unify the multivariate dynamic control charts by adjusting the warning limit and the sampling rule. Finally, an example of the application of VSI MAEWMA control chart is given, which can give some enlightenment to the practical engineering application.