Some Efficient Control Charts For Monitoring Process Location

Statistical process control(SPC)is a collection of statistical procedures designed to ensure that a manufactured item or performed service adheres to a defined set of quality criteria or meets the requirements of the clients.A control chart is a powerful online process monitoring tool of SPC being widely used to distinguish between natural and assignable causes of variations.The development of efficient design structure of control charts is always desirable for improved process monitoring.This thesis presents some new efficient control charts for improved process monitoring under different scenarios that will be helpful for quality practitioners to enhance the quality of the products in many industrial processes.In situations where some supplementary information related to the quality characteristic being monitored is available,we can use it advantageously to enhance the efficiency of the charting statistics and thus increase the sensitivity of the control charts.In this context,for monitoring the location of normal processes,we have incorporated auxiliary information into the CUSUM charting structure by using the ratio-type estimators as a substitute of the simple mean estimator.In Shewhart type control charts,the decision about the state of a process is made by using the information of a single sample which results in delayed small process shift detection.Moreover,it may also be difficult for engineers to decide about the process state based on the information contained in a single sample.In this framework,we have proposed the generalized multiple dependent state(GMDS)sampling scheme which is an extension of multiple dependent state(MDS)scheme.This proposed charting strategy consists of two pair of control limits and encompasses the Shewhart and MDS schemes.We have integrated GMDS scheme to design a variable control chart for effective monitoring of the location of normal processes.Moreover,we have developed a multivariate attribute control chart under GMDS scheme for the situations where the focus is on the number of defects in each item,and these defects may be classified into several categories.In this study,we have also considered the scenario where the measurement process physically damages the product,or it is based on a single observation per time period.In this framework,we have proposed two optimal synthetic Tukey's control charts for efficient monitoring of the process mean by integrating the confirming run length chart into the Tukey's control charts.These charts are more flexible and robust to outliers.We have also investigated the situation when the underlying process distribution is either unknown or non-normal.In this context,we have designed two efficient distribution-free homogeneously weighted moving average control charts.These charts are based on the sign and sign-ranked statistics for monitoring the process location under asymmetrical and symmetrical distributions,respectively.The performance of the proposed control charts is evaluated by using various run length characteristics.We have also provided some empirical illustrations for the practical implementation of the proposals included in this study.