Improving the power of out of control states detection in multivariate processes using variable sampling intervals policy

The familiar multivariate process monitoring and control procedure is the Hotelling’s T2 control chart, a direct analog of the univariate shewhart chart. But, its efficiency for detecting small to moderate shifts in the process mean is poor. To improve the power of chart, this paper presents the variable sampling intervals (VSI) scheme. It is assumed that the length of time the process remains in control has exponential distribution. The chart is modeled using Markov chains and is optimized using genetic algorithm optimization method. The results show that the T2 chart with variable ratio sampling scheme is quicker than the classical one in detecting almost all shifts in the process mean