Abstract: Aiming at the problem that it was difficult to balance the diversity and convergence of multi-objective particle swarm optimization algorithms, we proposed a dynamic population-based multi-objective particle swarm optimization algorithm. The increase or decrease of the population size of this algorithm depended on the resources in the archive, thereby  regulating the population size. On the one hand, particles were added by local perturbation based on grid technology to increase the local search ability of particles and improve the diversity of the algorithm. On the other hand, in order to prevent the population size from overgrowing, non-dominated ordering and population density were used to control the population size and  accelerate the algorithm search progress, avoiding premature convergence. Five comparative algorithms were selected for experiments on test functions, and the experimental results show that this algorithm has obvious diversity and convergence advantages.

Key words: dynamic population, particle swarm optimization, multi-objective optimization, diversity, convergence