Hybrid PSO Algorithm with Adaptive Step Search in Noisy and Noise-free Environments

Abstract

Particle Swarm optimizer (PSO) is a population-based algorithm applied to many applications due to its competitive performance. As a pioneering variant, Dual-Environmental Particle Swarm optimizer (DEPSO) solves optimization problem- s both in noisy and noise-free environments. This paper employs Adaptive step search (ASS) as an improvement of DEPSO by enhancing the information utilization. ASS is an efficient scheme to solve the stochastic point location (SPL) problem. It magnifies or shrinks the step size of a learning mechanism (LM) adaptively according to historical success or failure. This method allows each particle to search with an adaptive step size by enhancing the utilization of historical information. Experimental results performed on CEC2013 benchmark functions indicate that DEPSO-ASS outperforms DEPSO and other the state-of-art PSO variants in both noise-free and noisy environments.