Cooperative bare bone particle swarm optimization

Although bare bone particle swarm optimization (BPSO) is a promising algorithm without employing accelerating coefficients compared with traditional particle swarm optimization (PSO), it also inevitably tends to converges prematurely, especially for problems with multiple extremes. In this paper a cooperative learning strategy is applied to enhance the performance of BPSO. The proposed method uses a group of particles participating in exploring optimal solution. Depending on how the particles contribute to search step size, three different versions have been proposed and tested. The performances, both in solution quality and convergent rate, of these algorithms will be reported here. Tested on a suite of unimodal and multimodal benchmark functions justified the feasibility of the proposed strategy.