Change point detection and control of automatic continuous processes

Statistical process control (SPC) and automatic process control (APC) are two of the most commonly used on-line process control techniques. Both techniques minimize the deviation of the process from its target value. However, they are implemented in two entirely different ways. While SPC monitors the process using statistical methods, APC makes continuous corrective actions.;In the context of SPC, this dissertation introduces a so called cross-correlation control chart which is designed to detect linear shifts in the mean of a univariate process. The cross-correlation chart utilizes the average of observations in a given sampling period and the slope of the line fitted to these observations. The underlying distribution of the cross-correlation statistic is derived when the process is in control. Research results that compare the performance of the cross-correlation chart to appropriately designed X¯-trend chart are presented. We then extend the application of the crosscorrelation chart to monitor bivariate processes. Its performance is compared with those of the MCUSUM and Hotelling's T2 charts. Furthermore, a methodology is developed to calculate the control limits of the cross-correlation control chart for a given Type I error probability.;This dissertation provides the first realistic integration of SPC and APC through a hierarchical approach which is called statistical and automatic process control (SAPC). SAPC is designed for an automatically controlled process which is monitored by a control chart. Although the controller in such systems takes care of the disturbance to some extent, there might be situations where an intervention beyond the controller's actions is needed, such as in the case of equipment failure. Comparing the actual system with parallel systems that model the actual system under possible disturbances, SAPC detects and identifies the disturbance, and determines the time it started. Then a corrective action is applied either to minimize the deviation from the target or to remove the disturbance from the system. The application of SAPC is presented on an APC system consisting of three serially connected continuously stirred tank reactors and a PI controller under a step input disturbance.