Heuristic multi-objective optimization for cost function weights selection in finite states model predictive control

This research work investigates an automated and optimal procedure for the selection of the cost function weights in Finite States Model Predictive Control (FS-MPC). This is particularly useful where the cost function is composed by more variables and where other control parameters need to be carefully designed. A Genetic Algorithm (GA) multi-objective optimization approach is here proposed and tested on a case study represented by the FS-MPC of a Shunt Active Power Filter (SAF). The results of this weights optimization procedure are reported and discussed with the aid of Matlab-Simulink simulation tests.