Multi-loop PI tuning using a multiobjective genetic algorithm

In this paper, a method for the offline design of controllers is proposed to automate the design process. A multiobjective genetic algorithm is used to minimize the gap between the desired and the simulated percent overshoot and settling time. The Routh-Hurwitz criterion is incorporated into the design methodology to accelerate convergence by reducing the decision space. The proposed method is applied in a multiloop design for a DC motor. Simulation results show that the PI controller gains found using the proposed method give closed loop dynamics that follow the specified transient response. The results obtained using the proposed methods are shown to give better results than a conventional pole placement tuning method.