A Hybrid Evolutionary Algorithm by Combination of PSO and GA for Unconstrained and Constrained Optimization Problems

Abstract

A novel hybrid evolutionary algorithm (HEA) by combination of particle swarm optimization (PSO) and genetic algorithm (GA) is proposed for solving unconstrained and constrained optimization problems. In the proposed algorithm, evolution process is divided into two stages. In the first stage similar to PSO, particle flies in hyperspace and adjusts its velocity by following particles with better fitness according to flying experience of itself and its neighbors. In the second stage similar to GA, genetic operators of selection, reproduction, crossover, and mutation are exerted on particles at predetermined probability. Roulette-wheel selection operator selects particles with better fitness into next generation with more chance, single-point crossover operator shares better genetic schemes between particles, and Gaussian mutation operator gives particles opportunity of escaping from local optimum. By combination of PSO and GA, evolution process is accelerated by flying behavior and population diversity is enhanced by genetic mechanism. The proposed algorithm is tested on some standard unconstrained and constrained optimization functions. Satisfactory results obtained in the tests show that HEA can effectively balance searching ability of global exploitation and local exploration and is superior to PSO and GA in the solution of complex optimization problems.