Enhanced particle swarm optimization with multi-swarm and multi-velocity for optimizing high-dimensional problems

Traditional particle swarm optimization (PSO) algorithm mainly relies on the history optimal information to guide its optimization. However, when the traditional PSO algorithm searches high-dimensional complex problems, wrong position information of the best particles can easily cause the most of the particles move toward wrong space, so the traditional PSO algorithm is easily trapped into local optimum. To improve the optimization performance of the traditional PSO algorithm, an enhanced particle swarm optimization with multi-swarm and multi-velocity (MMPSO) is proposed. It comprises three particle swarms and three velocity update methods. The information sharing of the multi-swarm with various velocity update methods in the MMPSO can quickly discover more useful global information and local information, helping prevent particles from falling into local optimum and improving optimization precision of the algorithm. The MMPSO is tested on fourteen benchmark functions, and is compared with the other improved PSO algorithms. Comparison results validate the validity and feasibility of the MMPSO to optimize high-dimensional problems.