Adaptive Tuning of Model Predictive Control Parameters based on Analytical Results

Abstract

In dealing with model predictive controllers (MPC), controller tuning is a key designing step. Various tuning methods are proposed in the literature which can be categorized as heuristic, numerical and analytical methods. Among the available tuning methods, analytical approaches are more interesting and useful. This paper is based on a proposed analytical MPC tuning approach for plants which can be approximated by first-order plus dead-time models. The performance of such methods fails to deal with unknown or time-varying parameter plants. To overcome this problem, adaptive MPC tuning strategies are practical alternatives. The adaptive MPC tuning approach proposed in this paper is based on on-line identification and analytical tuning formulas. Simulation results are used to show the effectiveness of the proposed methodology. Also, a comparison of the proposed adaptive tuning method with a well-known online tuning method is presented briefly which shows the superiority of the proposed adaptive tuning method.