A novel particle swarm optimization based on prey-predator relationship

Particle swarm optimization (PSO) is a widely used nature-inspired optimization algorithm based on population and has strong robustness and good global astringency. In the mid-late iterations in the latest PSOs, there are plenty of dense gathering “slothful particles” with low velocities, which not only contribute little to the optimization but also impact the computation speed. Inspired by the prey-predator relationship in nature, we propose a novel prey-predator PSO (PP-PSO) that employs the three strategies of catch, escape, and breeding. In PP-PSO, slothful particles can be deleted or transformed, and while the former helps to speed up convergence and computation speed, the latter improves optimization results. In addition, a proportional-integral (PI) control is introduced for population control, where the population fluctuates but within a relative stability over iteration, thus enhancing population diversity. The experimental study on 10 basic benchmark functions and 30 advanced benchmark functions from CEC 2017 with different dimensions shows that our PP-PSO has a superior performance in comparison with ten other peer algorithms. In this study, a novel relationship between species behavior and swarm intelligence algorithm is found and the mechanism of particle motion in the convergence process of PSO is further revealed.