Síntese Estruturada para Controladores Preditivos LPV

Abstract

A novel framework for the design of dissipative Model Predictive Control (MPC) schemes for Linear Parameter Varying (LPV) systems is presented in this paper. This tool allows to take into account both time- and frequency-domain requirements, as well as optimal performance costs, directly into the synthesis procedure. The main novelty is to detach the parameter-dependency by using a stabilising LPV feedback that ensures dissipativity of a closed-loop storage function, as well as pole-placement requirements. By doing so, the main drawback of LPV MPC synthesis, the unavailability of the future values of the scheduling parameters, is avoided, since the MPC uses gain-scheduled linear predictions. Furthermore, the framework eliminates the need for stabilising terminal ingredients in the optimisation, since the LPV feedback ensures input-to-state stability by itself and the recursive feasibility of the MPC is derived from the dissipativity arguments. The proposed method is tested on a benchmark example and compared against a standard min/max MPC method. It exhibits good performance, obtained with reduced computational load since only one quadratic program is solved per sample.