In this thesis, an on-line quality improvement strategy combining automatic process control (APC) and statistical process control (SPC) is proposed. Run-to-run (RTR) control is a form of automatic process control (APC) to compensate process variation on a run-to-run (periodic) basis. One control strategy for RTR is the proportional-integral-derivative (PID) control schemes, which are used extensively for process control and variation reduction. Here the efficiency and robustness properties of RTR PID control schemes for some commonly encountered stationary and non-stationary stochastic disturbance processes are studied. Both theoretical and numerical results are obtained to show that the RTR PID schemes have reasonably high efficiency and are also robust, especially in the presence of non-stationarity.; Using SPC to monitor the special causes of the process along with APC is known to be an important tool of on-line quality improvement. Here the limitations of conventional SPC for the PID controlled processes are discussed. To overcome these limitations, a joint monitoring strategy to monitor both the controlled outputs and the manipulated inputs using bivariate SPC is proposed, and a specific design procedure taking into account both the controller and the disturbance model is provided. Also, the run length properties of the joint monitoring schemes using the Hotelling's approach and the Bonferroni's approach are systematically investigated. A new robustness index has been defined to investigate the robustness of the proposed monitoring schemes with respect to disturbance model uncertainty. Finally, a decision rule to select an appropriate monitoring scheme is proposed.; This research has significantly expanded the scope of conventional SPC research, and has great potential toward cultivating multi-disciplinary research activities between applied statistics and control engineering and their implementations in quality improvements in manufacturing.
展开▼
