Evaluation of a competitive particle swarm optimizer in multimodal functions with complexity

In this paper, we present a simple competitive particle swarm optimizer (CPSO) for finding plural solutions. In the CPSO, particles are divided into groups corresponding to the required number of solutions. Each group simultaneously searches solutions having a priority search region. This region affects to prohibit that different groups search the same solutions. The CPSO can effectively find desired plural acceptable solutions with a high accuracy and with a low computation cost, and can easily control combinations of these solutions by adjusting a parameter. This paper evaluates the CPSO in complex global optimization benchmarks. Through the numerical experiments, searching performances of the CPSO are clarified.