MODEL PREDICTIVE STATISTICAL PROCESS CONTROL - HANDLING STEP UPSETS

Abstract

A feedback-based model predictive control (MPC) approach to product quality improvement that incorporates multivariate statistical techniques has been developed [1]. The objective of the approach is to use existing process measurements to help reduce the variability of product quality when its online measurement is not feasible. The approach is model based and it uses principal component analysis to compress selected process measurements into scores. One or more manipulated setpoints are chosen and varied to control the scores in order to counteract the effect of stochastic process disturbances on product quality. The approach assumes that the selected process measurements correlate with product quality, and that the stochastic disturbances that cause product variability are stationary. When implemented on the Tennessee Eastman process the approach resulted in a 44% reduction in the variability of the product quality. In this paper the issue of how to handle non-stationary step upsets is addressed. A steady state model predictive control approach is used in conjunction with the dynamic score control to overcome the problems caused by the step disturbances.