A constrained model-based predictive control algorithm and its applications

This paper considers the direct computation of constrained model-based predictive controllers using dynamic programming. Process constraints (absolute and incremental limits on process actuation and measured variables) are formulated as linear inequalities in the constrained optimisation problem. Efficient methods have been developed for fast computation of the optimal control decisions. The application of the constrained controller to a real-time simulation of a BTX (benzene, toluene and xylene) separation process and a process involving nonlinearities is described.