The Research On Multi-objective Particle Swarm Optimization Algorithms Based On Decomposition Strategy

Particle swarm optimization algorithm has the advantages in solving multi-objective problems with simple encoding,easy to implement,fast convergence speed etc.Therefore, the particle swarm algorithm has become a research hotspot in multi objective optimization. But the approximate solution set of multi-objective particle swarm optimiza-tion algorithm should be improved in terms of distribution uniformity, diversity, convergence. Moreover, the efficiency of the alogrithm need further improvements.Based on multi-objective particle swarm optimization algorithm, decomposition strategy is regarded as a breakthrough point in our paper. In view of the problem of efficiency of particle swarm optimization algorithm and its approximate set solution uniform distribution, diversity, convergence, we made improvement and innovation. Two effective algorithms were proposed. The main work is as follows:(1)The classical Tchebycheff decomposition strategy is applied to the multi-objective particle swarm optimization algorithm, we propose a hybrid particle swarm algorithm for multi-objective optimization based on decomposition Smoeadpso.Based on the frame of the algorithm, we use a new global optimal value of the particle selection strategy, and propose a new particle flight speed limit strategy to effectively control the particle flight space, and use the elitist solution set retention strategies to retain the excellent solution set. Comparing with excellent and classic NSGAII, MOEA/D and multi-objective particle swarm optimization algorithm OMOPSO shows that the convergence of the algorithm and the diversity in two objective problems are better than the compared algorithms,but the uniformity of the solution set still should be improved.(2) We design a new multi-objective particle swarm optimization algorithm based on decomposition HEMOPSO. In the premise of maintaining excellent convergence of the Someadpso algorithm, HEMOPSO is further improved in the uniformity and efficiency. In order to improve the uniformity of the solution set, our algorithm use the penalty boundary intersection decompostion strategy which can create uniform set of solutions. In order to improve the efficiency of the algorithm, our algorithm proposes a simple and fast global optimal selection strategy. Based on the theory that the particles flight to the true Pareto front face, the algorithm abandon the external archive set strategy. In order to prevent the particle falling into the local optimum in the later period, the mechanism of particle aging is applied. By comparing with two multi-objective particle swarm optimization algrithm OMOPSO and Smoeadpso, we prove that the algorithm retains the good convergence,while the diversity, uniformity and efficiency of the solution set is improved to some extent.