Comparison of Evolutionary Multi-Objective Optimization Algorithms on the Tuning of PI Controllers for Electric Drives

This paper presents a comparison of the two most used evolutionary algorithms for multi-objective problems, Strength Pareto Evolutionary Algorithm 2 (SPEA2) and Non-Dominated Sorting Genetic Algorithm II (NSGA-II), applied to the position control of a DC motor drive. The DC motor drive is considered with three cascade-controlled closed loops: armature current, speed and position. The outputs of the PI armature current and speed controllers are limited to the rated armature voltage and current, respectively. Because of this, higher gains for the controllers can be used, resulting in faster responses. However, windup phenomenon can arise. To avoid this, anti-windup circuits are used and, consequently, the system becomes non-linear, making the optimal tuning of the controllers even more challenging. To address this problem, a couple of multi-objective evolutionary algorithms have been employed to tune all the controllers simultaneously, together with the anti-windup window. In this paper two of them, SPEA2 and NSGA-II are investigated. They are implemented in MATLAB, while the electric drive model was developed in the SIMULINK environment. Simulation results demonstrate the algorithm's effectiveness in achieving optimal tuning for all three PI controllers while satisfying multiple conflicting objectives. Statistical tests were carried out demonstrating the advantage of SPEA2 over NSGA-II.