Performance monitoring of MPCA-based control for multivariable batch control processes

Abstract On-line control loop performance monitoring is developed based on multiway principal component analysis (MPCA) and minimum variance outputs for batch operation processes. Unlike continuous processes, the performance assessment of batch processes requires particular attention to batch disturbances and set-point changes. Because of the intrinsically nonlinear dynamic behavior of batch processes, a linear time-variant system for batch processes is used here. The traditional MPCA-based monitoring can only differentiate between normal and abnormal process behaviors since the data used in constructing the control limits is based on the normal operating condition. It cannot indicate whether or not the process is at the optimal or suboptimal condition. In this paper, MPCA is modified so that it can also be used to determine if the process is operated at optimum conditions. This is done by building performance bounds based on data from batch runs operated at optimum conditions. The advantages of the proposed method are illustrated using an exothermic chemical batch reactor.