Particle Swarm Optimization Algorithm with Adaptive Chaos Perturbation

Most of the existing chaotic particle swarm optimization algorithms use logistic chaotic mapping. However, the chaotic sequence which is generated by the logistic chaotic mapping is not uniform enough. As a solution to this defect, this paper introduces the Anderson chaotic mapping to the chaotic particle swarm optimization, using it to initialize the position and velocity of the particle swarm. It self-adaptively controls the portion of particles to undergo chaos update through a change of the fitness variance. The numerical simulation results show that the convergence and global searching capability of the modified algorithm have been improved with the introduction of this mapping and it can efficiently avoid premature convergence.