A Self-organizing Multi-objective Particle Swarm Optimization Algorithm for Multimodal Multi-objective Problems

To solve the multimodal multi-objective optimization problems which may have two or more Pareto-optimal solutions with the same fitness value, a new multi-objective particle swarm optimizer with a self-organizing mechanism (SMPSO-MM) is proposed in this paper. First, the self-organizing map network is used to find the distribution structure of the population and build the neighborhood in the decision space. Second, the leaders are selected from the corresponding neighborhood. Meanwhile, the elite learning strategy is adopted to avoid premature convergence. Third, a non-dominated-sort method with special crowding distance is adopted to update the external archive. With the help of self-organizing mechanism, the solutions which are similar to each other can be mapped into the same neighborhood. In addition, the special crowding distance enables the algorithm to maintain multiple solutions in the decision space which may be very close in the objective space. SMPSO-MM is compared with other four multi-objective optimization algorithms. The experimental results show that the proposed algorithm is superior to the other four algorithms.