Balance Search Particle Swarm Optimization

Abstract

Particle swarm optimization approach is a swarm based procedure, used to search global optimum solution in a search domain. The Basic PSO algorithm show premature convergence to find global optimum solution (best fitness value of particle). The fitness function value may be maximum or minimum value depends on application. Relative search behaviour of PSO particles are to be guided according to the fitness value of gbest particle. This paper proposed an approach which increases global exploration and local exploitation ability of basic PSO algorithm using angular relation between solutions. The proposed approach named as Balance Search Particle Swarm optimization (BS-PSO) model search space as a concentric circular search environment. The BS-PSO algorithm maintains optimum search behavior, from inception of the search procedure to wider search space. Initially BS-PSO approach provide a controlled and directed procedure to find optimal solution using search exploitation and later on it increase exploration ability of the search procedure to capture global optimum solution. In this manner BS-PSO approach maintain a perfect diversity among solution and prevent PSO algorithm to trap in local optima and premature convergence situation.