Research On CUSUM Control Chart Of The Auto-correlated Process And Its Application

In the production processes, the control chart technique is an effective tool for quality assurance, and is widely used in the world. When autocorrelation exists in the production process, the Shewhart control chart is often disabled, and its detection performance has been greatly affected, resulting in human, financial and material resources of a great waste. With the increasing popularity of automation equipment, automatic sampling in continuous production becomes a reality; therefore, autocorrelation will be increasingly common. Autocorrelation process is concerned by the international quality management experts, and there are a series of research achievements, and in research methods the experts focus on adjusting the control limits, time series models, and exponentially weighted moving average control chart for monitoring the auto-correlated process.This paper refers to international quality management techniques and methods, and studies the cumulative sum control chart of the auto-correlated process systematically. First, by analyzing the characteristics of the auto-correlated process, this paper discusses the methods of eliminating autocorrelation in the process, and also analyzes the modified X control chart and the residual control chart. Second, by using cumulative sum methods of the traditional process and time series model, this paper builds the CUSUM control charts for monitoring the auto-correlated process, and provides the implementation methods of CUSUM control chart for monitoring the autocorrelation process. Meanwhile, a comparison about the average run lengths of the CUSUM charts and the modified X charts for monitoring the auto-correlated process is presented, and the result shows that the CUSUM control chart is more sensitive in the auto-correlated process. Finally, this paper conducts the simulation analysis for the autocorrelation process of drying the cut tobacco in the cigarette production, and comparatively studies the application effect of the CUSUM chart and the residual control chart for monitoring the auto-correlated process, and also further verifies that the CUSUM chart for monitoring the auto-correlated processes is more effective in finding the small fluctuations of the process mean.This study's features and innovations are mainly presented:first, the control limits are modified in the X control chart for monitoring the autocorrelation process,while this paper fits the process model and seeks the residuals used in the CUSUM control chart, and it can effectively eliminate the autocorrelation; second, this paper constructs the CUSUM chart in the auto-correlated process, and compares with the modified X control chart and the residual chart; third, this paper provides the application effect of the CUSUM chart and the residual control chart by the empirical analysis of drying the cut tobacco.This paper not only provides the principles and methods of the CUSUM control charts for monitoring the auto-correlated process, and also affords the theoretical basis and implementation approach to other industries by empirical research on the tobacco industry.