Multi-Objective Gain Optimizer for an Active Disturbance Rejection Controller

Abstract

Active Disturbance Rejection Control (ADRC) has proven to be an efficient control method, however, the tuning of its parameters is a complicated endeavor. This paper explores the use of reference point based dominance in the traditional multi-objective non-dominated sorting genetic algorithm (NSGA-II) to perform the parameter tuning. The algorithm is applied to a simulation and physical implementation of an inverted pendulum system. The optimization method generated values that offered suitable performance among various fronts.