Particle swarm optimization based on particle perturbation and elite preservation strategies for dynamic optimization problems

Multi-population methods have an effective ability to solve various dynamic optimization problems. However, most multi-population methods still have much room for improvement in their global search capabilities. To address this issue, a particle swarm optimization based on particle perturbation and elite preservation strategies (PSO-PE) is designed to solve dynamic optimization problems. The particle perturbation strategy is first designed to enable the optimal particles of each explorer sub-population to escape from local optima. Secondly, the elite preservation strategy is proposed to prevent multiple sub-populations from converging on the same peak by preserving elite particles while excluding others. Finally, an elite particle migration strategy is proposed to cope with environmental changes. Specifically, the exploiter sub-populations can be used to trace the optimal solution in a new environment while the explorer sub-populations are adopted to maintain the diversity. We conduct the experiment on the Moving Peak Benchmark (MPB) and Generalized Moving Peaks Benchmark (GMPB) problems. The comprehensive results have demonstrated the superior performance of the proposed method compared with some related dynamic optimization algorithms.