Adaptive Inertia Weight Particle Swarm Optimization

Abstract

Adaptive inertia weight is proposed to rationally balance the global exploration and local exploitation abilities for particle swarm optimization. The resulting algorithm is called adaptive inertia weight particle swarm optimization algorithm (AIW-PSO) where a simple and effective measure, individual search ability (ISA), is defined to indicate whether each particle lacks global exploration or local exploitation abilities in each dimension. A transform function is employed to dynamically calculate the values of inertia weight according to ISA. In each iteration during the run, every particle can choose appropriate inertia weight along every dimension of search space according to its own situation. By this fine strategy of dynamically adjusting inertia weight, the performance of PSO algorithm could be improved. In order to demonstrate the effectiveness of AIW-PSO, comprehensive experiments were conducted on three well-known benchmark functions with 10, 20, and 30 dimensions. AIW-PSO was compared with linearly decreasing inertia weight PSO, fuzzy adaptive inertia weight PSO and random number inertia weight PSO. Experimental results show that AIW-PSO achieves good performance and outperforms other algorithms.