Design Of Three Adaptive Nonparametric Control Charts For Monitoring Scale Parameter

Control charts are one of the important tools for statistical process control,which can be directly used to control and diagnose production processes.They play major roles in improving production efficiency.At present,conventional parameter control charts are often designed for common distributions or a specific distribution.But the distribution that the production process obeys is usually unknown in many manufacturing processes.when the product quality obeys other distributions,control charts have poor ability.The nonparametric control charts solve this problem very well.It does not require the producer to know the distribution of the process in advance.It has good detection performance under different types of distributions.However,most control charts are used to monitor shifts of the process mean.The dispersion charts are rare.As one of the important indicators for judging the fluctuations of the process,the establishment of an effective dispersion control chart can help producer judge whether the product quality is stable.In addition,the detection capability of traditional CUSUM and EWMA charts often depends on the shift size specified in advance.In production,the actual amount of shift that occurs out of control of the process is often uncertain.In order to overcome these problems,we propose the adaptive non-parametric EWMA chart with score function,the adaptive non-parametric EWMA chart with shift estimation function and the Nonparametric CUSUM chart with variable sampling interval,respectively.These charts use to monitor shifts in the process variability.Through Monte Carlo simulations,the control limits and average running length or average time of sign are solved.We explore the influence of different parameter combinations on the dispersion charts and the performance of control charts under in-control and out-of-control conditions.Finally,the proposed control charts are applied to the examples.Finally,the proposed control charts are applied to the real examples.The results show that the proposed non-parametric control charts based on adaptive method for monitoring shifts in the process variability have better detection performance.The proposed charts are more effective than that of conventional non-parametric control charts in detecting small to large upward shifts.