Tracking Multiple Optima in Dynamic Environments by Quantum-Behavior Particle Swarm Using Speciation

摘要

This paper presents an improved Quantum-behaved Particle Swarm Optimization, namely the Species-Based QPSO (SQPSO), using the notion of species for solving optimization problems with multiple peaks from the complex dynamic environments. In the proposed SQPSO algorithm, the swarm population is divided into species (subpopulations) based on their similarities. Each species is grouped arounda dominating particle called species seed. Over successive iterations, species are able to simultaneously optimize towards multiple optima by using the QPSO procedure, so that each of the peaks can be definitely searched in parallel, regardless of whether they are global or local optima. A number of experiments are performed to test the performance of the SQPSO algorithm. The environment used in the experiments is generated by Dynamic Function # 1 (DF1). The experimental results show that the SQPSO is more adaptive than the Species-Based Particle Swarm Optimizer (SPSO) in dealing with multimodal optimization in dynamic environments.