Particle Swarm Optimization with a Novel Multi-Parent Crossover Operator

Particle swarm optimization (PSO) shares many similarities with evolutionary algorithms (EAs), while the standard PSO does not use any evolution operators such as crossover and mutation. This paper presents a hybrid PSO algorithm to inherit some excellent characteristics of advanced evolutionary computation techniques. The proposed method employs a novel multi-parent crossover operator and a self-adaptive Cauchy mutation operator to help escape from local optima. Experimental results on a suit of well-known benchmark functions with many local minima have shown that the proposed method could successfully deal with those difficult multimodal optimization problems.