Research Of Nonparametric Statistical Quality Control Chart Based On Cucconi Test And Change-Point Model

Statistical quality control chart is a very primary tool for controlling and improving quality, which is an important strategy for industry organizations to delight customers. But traditional control charts have poor robustness, and couldn’t make full use of data information, which can result in slow and inaccurate response to abnormal condition of production process. Therefore, it’s an urgent need of developing more efficient and more robust control charts.Supported by the NC generation program of Ministry of Science and Technology "Zhejiang NC generation mechanical product innovation demonstration project ", this paper proposes a nonparametric statistical control chart formed by integrating the Cucconi test into the widely researched change-point model framework, which can quickly detect changes in both location parameter and scale parameter, and diagnose the type of exception accurately.Firstly, the background and significance of this research is introduced. The research status of Shewhart control chart, EWMA control chart and CUSUM control chart is detailed, and also some other nonparametric control charts that derived from these three kinds of control charts. In addition, the change-point model and some control charts based on it are presented.Secondly, after studying the theory of statistical quality control chart, change-point model and the property of rank statistic, a nonparametric statistical control chart based on the Cucconi test and change-point model is established, (CCNSPC Chart for short in this paper). For CCNSPC Chart, the Monte Carlo simulation method is selected to find out the control limits, and a fitting function is presented for users’convenience.Thirdly, in order to measure performance of CCNSPC chart as the average time taken to detect the change(ARL1), CCNSPC chart is compare with three other charts of the same type by Monte Carlo simulation method. Simulation study suggests that the CCNSPC chart gives a best performance of all. Furthermore, experiment suggests a combination of Mann-Whitney test and Ansari-Bradley test to judging exception types, after estimating the change point.Finally, the CCNSPC chart is applied on an automatic production line of motor cover, from which a conclusion is made that the CCNSPC chart has fast response to the abnormal process, accurate judgment of the position of change point, and can judge the type of change precisely.