Since optimizing the values of parameters in Particle Swarm Optimization (PSO) can improve its performance of finding a globally optimal solution, meta-optimization methods have received wide attention in computer science and various application disciplines. This paper proposes Evolutionary Particle Swarm Optimization (EPSO) that can systematically estimate appropriate values of parameters in PSO for a given optimization problem by genetic computation without prior knowledge. To demonstrate the effectiveness of EPSO, computer experiments on benchmark problems are carried out. We give experimental results and analyze the features of EPSO.
展开▼