Particle swarm optimization based on power mutation
Xiaoling Wu, Min Zhong · 2009
Particle swarm optimization (PSO) has shown its fast search speed and good search ability in many optimization problems. However, PSO easily suffers from local minima when dealing with complex problems. In order to enhance the standard PSO, this paper presents an improved PSO algorithm, namely PMPSO, which employs a power mutation (PM). The main idea of PMPSO is to conduct a PM on the global best particle in current swarm. It is to hope that the mutation could help particles jump out local optima. To verify the performance of the proposed approach, PMPSO is compared with some existing algorithms, PSO with Cauchy mutation (HPSO), PSO with near neighbor interactions algorithm (FDR-PSO), classical evolutionary programming (CEP), and fast evolutionary programming (FEP) on ten well-known benchmark functions. Experimental results show that PMPSO achieves better results on majority of test functions.