Hybridizing multi-objective, clustering and particle swarm optimization for multimodal optimization

Multimodal optimization problems (MMOPs) are a kind of common optimization problem aiming to find multiple high-accurate optimal solutions. In this paper, a multimodal optimization algorithm named MO-C-PSO is proposed. In the proposed method, we combine the exploration ability in the whole search space of multi-objective technique with the special clustering ability of mean-shift clustering method. In this way, each potential optimal individual is induced to form its own unique sub-population. Then, particle swarm optimization (PSO) guides the local search in each sub-population by applying the proposed switching evolutionary process (SEP) strategy, which can refine the solution accuracy. To evaluate the performance, the proposed method is compared with other state-of-the-art methods on CEC'2013 benchmark set, the classic high-dimensional problems, and a real-world application. The experimental results have validated that MO-C-PSO can provide competitive performance.

Automated summary

The MO-C-PSO algorithm combines the exploration ability of multi-objective technique with the clustering ability of mean-shift clustering method, providing competitive performance.