Genetic-Algorithm-Based Optimization of Ant Colony Controller for Fractional-Order Systems

Abstract

Finding appropriate initial parameter values for Ant Colony Optimization (ACO) is a research gap that has limited usage of ACO as an optimization technique. Conventionally, parameters for ACO are initiated by a trial and error procedure. We propose to use Genetic Algorithm (GA) for finding optimal initial ACO parameters for control of fractional-order systems. This leads to an efficient and reliable ACO controller with quicker convergence. Proposed GA-ACO approach uses a nested GA with ACO for fractional PID controller tuning by minimizing a multi-objective function. We simulate our proposed approach on five different fractional-order systems and compare its performance against: (a) ACO based and (b) GA-based fractional-order controller. Simulation results show that our GA-ACO approach outclasses ACO and GA fractional controllers in terms of better transient response (rise time, peak overshoot) and steady-state response (settling time, ITAE), with a higher computational complexity.