Ladder particle swarm optimization and application in function optimization

An adaptive variable population method was proposed. The population size was modulated based on diversity function in type of ladder, every particle was evaluated with adequate time and diversity of swarm was maintained, so the probability of local convergence was declined in large degree. Diversity of population plays a very important role in evolution, so in this method, a number of particles with small fitness would be deleted when the size of current population should be reduced, and crossover operator is used to increase a number of new particles into swarm, the new swam inherits the old swarm and the diversity of population is maintained. The effectiveness of the given algorithm is indicated by experiment of typical test function.