Improved centripetal accelerated particle swarm optimization

Abstract

Particle swarm optimization (PSO) is a bio-inspired optimization algorithm that imitates the social behavior of bird flocking, fish schooling and swarm theory. Although PSO has a simple concept and is easy to implement, it may converge prematurely because of its poor exploration when solving complex multimodal problems. Centripetal accelerated particle swarm optimization (CAPSO) is an enhanced particle swarm optimization (PSO) scheme combined with Newton's laws of motion. It has shown an effective algorithm in solving optimization problems however; its performance can be enhanced similar to other evolutionary computation algorithms. In this study, an improved CAPSO (ICAPSO) and improved binary CAPSO (IBCAPSO) are introduced to accelerate the learning procedure and convergence rate of optimization problems in the real and binary search spaces. The proposed algorithms are evaluated by twenty high-dimensional complex benchmark functions. The results showed that the methods substantially enhance the performance of the CAPSO for both the real and binary search spaces in terms of the convergence speed, global optimality, and solution accuracy.