The Change Point Identification Of Poisson Process Based On The Generalized Likelihood Ratio

Statistical process control is playing an increasingly important role in the processcontrol as a important tool in the field of quality management, After the developmentof the past few decades, theoretical research system of SPC has reached a new height.The problem of change-point identification in statistical process control is alsogradually shining onto the stage because of its important position. From1954-changepoint problem was put forward firstly, more and more scholars studied a variety ofchange point type and had made a lot of recognition of the change point method, but aproblem is that most of the change point identification method in previous studies isbased on the Normal distribution assumption, but little for Poisson distributionassumption. And the modification accuracy and technical aspects of the change pointalso need to be further improved. In this paper, based on the generalized likelihoodratio (GLR) method, a GLR-based model for identifying the change point in Poissonprocesses is proposed, the method is the expansion of the GLR application in normaldistribution assumption, and the effect on the detection accuracy is also verysignificant. In this paper, firstly, we introduce the theoretical background of themethod which will be used in this article, In the third chapter, we give the GLRchange point identification model including the initial parameter is known orunknown, the number of change points in process is one and two, and in the last of thethird chapter, we give the GLR model of multiple change points modification. In thefourth Chapter of this article, we give simulation experiment using Matlab, thisexperiment will propose the performance and reliability of the change pointidentification in Poisson process. Another innovative point of this paper is that wecombine the change point identification process with the seeking for the in-controldata, from the simulation experiment in chapter four, We verify that the GLR modelcan also simultaneously get process in-control data which can be used to estimate theprocess parameters in phase II. At the end of the article, we state the advantages anddisadvantages of this paper, give the technical summary and future research directionsin the change point field and Prospects.