Gaussian Function-Based Particle Swarm Optimization

Abstract

This paper presents the Gaussian function-based particle swarm optimization (PSO) algorithm. In canonical PSO, potential solutions, called particles, are randomly initialized in the beginning. The proposed method uses the solutions of another evolutionary computation technique called genetic algorithm (GA) for initializing the particles in order to provide feasible solutions to start the algorithm. The method replaces the random component of the velocity update equation of PSO with the Gaussian membership function. The Gaussian function-based PSO is applied on eight benchmark functions of optimization and the results show that the proposed method achieves the same quality solution in significantly fewer fitness evaluations. This proposed modification of PSO will be useful to optimize efficiently.