Research And Application Of Nonparametric Shewhart-Type Control Chart Based On Two-Sample Statistics

Control chart is the main tool to study statistical process control,which can be used for modern industrial quality monitoring.For the monitoring of metadata,Shewhart control chart,CUSUM control chart and EWMA control chart are generally used.The mon-itoring object is mainly a quality feature.For the monitoring of multivariate data,Hoteling~2control chart and EWMA multivariate control chart are generally used.However,the tra-ditional multivariate control chart often has the following two problems:1.It is assumed that the observations are independent of each other and follow the same normal distribution.2.Da-ta preprocessing by dimensionality reduction.In practical application,it is difficult to meet the assumption that the samples follow normal distribution,and dimensionality reduction will cause the loss of information to a certain extent.Therefore,aiming at the above two prob-lems,this paper proposes a nonparametric Shewhart-type control chart suitable for monitoring high-dimensional data.The control chart is easy to understand and operate.This paper has done the following work for the high-dimensional nonparametric Shewhart control chart:(1)In traditional methods,scholars often combine data depth and control chart to monitor whether multivariate high-dimensional data follow the same distribution,but the amount of cal-culation is too large.In this paper,we use Euclidean distance to transform multivariate data into metadata,and then combine Shewhart control charts with Wilcoxon statistic,Ansari-Bradley statistic and Lepage statistic to obtain three nonparametric Shewhart-type control charts based on two?sample statistics.(2)In this paper,the robustness of nonparametric Shewhart-type control chart is verified by Monte Carlo numerical simulation.Based on different distributions,sample sizes and different dimensions,we compare the monitoring effect of nonparametric Shewhart-type control chart when the location parameters or scale parameters of the distribution drift.(3)In this paper,the nonparametric Shewhart-type control chart is used to monitor the distribution of meteorological data in Nyingchi,Tibet in the first 100 days of 2019,and whether there is a difference compared with the first 100 days of 2018...