ON SOME ENHANCEMENTS TO MULTIVARIATE CONTROL CHARTS TO MONITOR PROCESS PARAMETERS

Multivariate control charts are often used to monitor production processes that consist of several correlated quality characteristics. The most widely used multivariate process monitoring tool is the Hoteling T-square control chart. The scheme is well-known for its effectiveness in separating common and special causes on variations. However, like with any other Shewhart control chart, the classical Hoteling T-square chart is not very effective in handling small shifts in a process, since it uses only information contained in the most-recent observation. In this research project, we propose to investigate new T-square or Chi-square, MCUSUM and MEWMA control chart using auxiliary information and adaptive schemes to increase the sensitivity of the chart to monitor the parameters of multivariate processes. Other efficient techniques such as the run-rules profiles, fast initial response and effective sampling plan will also be implemented in the proposed design structures. Using Monte Carlo Simulation procedure, we will analyse the detection ability of the new multivariate charts using different run-length characteristics as well as overall measures such as the extra quadratic loss (EQL), the average ratio for the average run-length (ARARL) among others. The performance of these proposed charts will also be compared to some other existing counterparts. Furthermore, practical application of these proposals will be demonstrated using real data from different disciplines such as in health care, analytical laboratories and production industries.