dynamic optimization algorithms is to continuously locate and track changing optimal solutions using limited computational resources. Hence, how to strengthen the exploration ability for locating the optimum of the static problem in an environment and how to improve the adaptation ability to the changing optima in different environments are two key issues for efficiently solving DOP. To address these issues, we propose a diversity-driven multi-population particle swarm optimization (DMPSO) algorithm. First, we propose a center inf strategy is proposed to reinitialize the population. Experimental studies are conducted on the moving peaks benchmark to compare the DMPSO algorithm with some state-of-the-art dynamic optimization algorithms. The experimental results show that the proposed DMPSO algorithm outperforms the contender algorithms which validate the effectiveness of the proposed algorithm.