Multi-objective particle swarm optimization with dynamic population size

There are many complex multi-objective optimization problems in the real world, which are difficult to solve using traditional optimization methods. Multi-objective particle swarm optimization is one of the effective algorithms to solve such problems. This paper proposes a multi-objective particle swarm optimization with dynamic population size (D-MOPSO), which helps to compensate for the lack of convergence and diversity brought by particle swarm optimization, and makes full use of the existing resources in the search process. In D-MOPSO, population size increases or decreases depending on the resources in the archive, thereby regulating population size. On the one hand, particles are added according to local perturbations to improve particle exploration. On the other hand, the non-dominated sorting and population density are used to control the population size to prevent the excessive growth of population size. Finally, the algorithm is compared with 13 competing multi-objective optimization algorithms on four series of benchmark problems. The results show that the proposed algorithm has advantages in solving different benchmark problems.