Design of an Adaptive Predictive Control Strategy for a MIMO Process

Abstract

An adaptive predictive (AP) control strategy for a multi-input multi-output (MIMO) industrial process is designed. The Relative Gain Array pairing technique is used to define the adaptive predictive control strategy, i.e. the AP models structure. The control strategy is simulated using the ADEX. The controllers' performance is evaluated on a simulator of a crude oil atmospheric distillation process which is currently operating in a Petroleos Mexicanos refinery. The simulation results are also compared with a PID-based control strategy, showing an improvement of operation stability.