IMPROVED MODEL PREDICTIVE CONTROL AND ITS APPLICATION

This paper presents a modified form of model based predictive control that exploits the concept of non-trivial terminal state weighting in the cost function. While maintaining simplicity of implementation, the performance of this algorithm is not sacrificed and the computational burden compared with traditional approaches is greatly reduced. This paper describes the proposed algorithm in detail and through application to several systems, including a benchmark fluidised catalytic cracker unit, demonstrates its advantages over traditional model based predictive control algorithms.