Crossover particle swarm optimization with incremental learning

Particle Swarm Optimization (PSO) algorithm is easy to fall into premature local convergence. In this paper, inheriting the crossover particle swarm optimization combined with the crossover operator, we proposed Crossover Particle Swarm Optimization with Incremental Learning (ILCPSO). ILCPSO builds two levels to overcome the local convergence and obtain the optimal solution. Firstly we introduce cross-operation into PSO, which exchanges good genes of particles in population according to certain crossover probability. It takes full advantage of information of the particle swarm to get the global optimal solution. Followed by the introduction of incremental learning, we add two sets of particles in prior to choose their own first teacher, then according to the diversity of the population use two different learning methods to update their knowledge again. After competition with the respective first teacher the fittest will be survival, in order to ensure that the population size will not expand. Final validation as opposed to the crossover particle swarm optimization algorithm, the complexity of the algorithm has not increased significantly when the performance is greatly improved. Copyright © 2013 Binary Information Press.