Control Chart For The Maxwell Distribution

The conventional control charts are based on the assumption that the distribution of the process to be monitored follows the normal distribution.However,in many practical situation in real life,this normality assumption may be violated.For example,the Maxwell distribution may be one of such candidate distribution.The performance of control chart will be influenced by using normal distribution designing scheme,when the real ditribution is skewed.In this paper,we propose two kinds of effective control charts for morning Maxwell distribution.One kind of charts are based on the maximum likelihood estimation of Maxwell scale parameter,inducing the one-side exponentially weighted moving average chart(EWMA),one-side cumulated sum chart(CUSUM)and double exponentially weighted moving average chart(DEWMA).The other kind of charts are based on the inverse normal transformation.The performance of the new charts are evaluated in terms of average run length(ARL)and standard deviation of run length(SDRL).The comparation betwen our new charts and other two existing charts is given.The simulation results show that the new charts are more effective than the other charts when the process undergoes small and moderate shifts.