A Novel Approach to Constrained Generic Model Control Based on Quadratic Programming

Abstract

The ability to handle constraints is a very important feature for process control algorithms. The conventional generic model control (GMC) uses general nonlinear programming to handle the constraints, which limits its industrial implementation. In this paper, after introducing an approximate model-based predictor, we present a quadratic programming-based optimization algorithm, which has the ability to handle linear constraints of manipulated and controlled variables and their moving velocities. By combination of the proposed optimization algorithm with the generic model control scheme, a novel approach to constrained generic model control based on quadratic programming is proposed for nonlinear affine systems with relative order 1. Computer simulation results show that the proposed approach has definite robustness against process/model parameter mismatches, it can be applied in real time, and it appears to hold a considerable promise in process control.