A Particle Swarm Optimizer with adaptive dynamic neighborhood for multimodal multi-objective optimization

In this paper, a multi-objective particle swarm optimizer based on adaptive dynamic neighborhood (ADN-MOPSO) is proposed to locate multiple Pareto optimal solutions to solve multimodal multi-objective problems. In the proposed algorithm, a spatial distance-based non-overlapping ring topology is used to form multiple subpopulations for parallel search to enhance the local search capability of the algorithm. In addition, an adaptive dynamic neighborhood selection strategy is proposed to balance the exploration and exploitation capabilities of the algorithm, allowing the size of the subpopulation to change automatically when the neighborhood switch time is met. To prevent the algorithm from premature convergence, a stagnation detection strategy is introduced to apply a Gaussian perturbation operation to the particles that fall into the neighborhood optimum. Finally, the proposed algorithm is used to solve multimodal multi-objective test problems and compared with existing multimodal multi-objective optimization algorithms. The results show that the proposed algorithm can obtain more Pareto solutions when solving different types of multimodal multi-objective functions.