Optimization Of Several Coefficients Of Variation Control Charts

As one of the key tools for statistical process control,control charts are usually used to monitor the average or variance of a process in traditional quality control applications.However,in some cases,owing to the mean or variance in some processes is not a constant,while the coefficient of variation(CV)is a constant when the process is controlled,therefore,it is more reasonable and more in line with the real situation to monitor the production process through the coefficient of variation.In order to enhance the monitoring efficiency of the coefficient of variation control chart,this dissertation proposes several ways to improve the coefficient of variation control chart.Firstly,this dissertation designs an unbiased CV control chart of the average time to sinal(ATS)under the condition that the sampling interval is variable,and derives the calculation formula of ATS,the control limit coefficient and warning limit coefficient of the control chart.By comparing the CV control chart of the fixed sampling interval and the variable sampling interval though ATS values,the result shows that the monitoring efficiency of the variable sampling interval is higher.Secondly,this dissertation applies the idea of cumulative sum to the CV control chart and designs a one-sided cumulative sum CV chart.The one-step transition probability matrix and the average run length(ARL)of the control chart is obtained by means of the idea of Markov chain.In addition,this dissertation also applies the Markov chain method to the one-sided exponentially weighted moving average CV control chart.By comparing and analyzing the efficiency of the two control charts on the production process,it is concluded that the cumulative sum CV chart is more efficient when the process CV has a slightly deviation.Finally,this dissertation combines the ideas of cumulative sum and exponentially weighted moving average to the same control charts to design two single-sided mixed CV control chart.By comparing and analyzing the cumulative sum CV diagram and two mixed CV diagrams in a random simulation manner,it is obtained that the mixed exponentially weighted moving average and the cumulative sum CV diagram have the highest monitoring efficiency when the process has a slightly deviation,while the cumulative sum CV diagram has a better performance when the process has a large deviation.