Chaos Particle Swarm Optimization Algorithm for Multi-Objective Constrained Optimization Problems

摘要

In this paper, an improved particle swarm algorithm based on chaos for multi-objective optimization problems is proposed. The algorithm uses external archive to save the current best solutions, and maintains the archive by density distance, the object dynamic crowding is used to measure Pareto solutions' quality and the strategy of eliminating low dynamic crowding solutions is employed to enhance non-dominated solutions uniformity. In order to preserve population diversity, the global best is selected randomly from those non-dominated solutions which have bigger density distance in the archive. Personal best position is updated by Pareto dominance relationship. When the basic particle swarm optimization gets into local convergence, chaos disturbance is introduced to guide the swarm to escape from local optima, so it can overcome the defect of traditional particle swarm optimization algorithm on getting into local best and enhance the global exploratory capability of PSO. The proposed algorithm is compared with two well known multi-objective evolutionary algorithms through three standard test functions, the numerical experiment results demonstrate that the obtained Pareto optimal solutions by the algorithm can rapidly converge to the Pareto front and uniformly spread along the front.