The current multi-objective particle swarm algorithms excel in convergence speed for solving complex problems but often suffer from a loss of population diversity. Conversely, composite differential evolution algorithms maintain superior solution distribution but lag in convergence efficiency. This research introduces an improved hybrid algorithm, CoDE-MOPSO, which integrates multi-objective particle swarm optimization with composite differential evolution based on clustering technology. The clustering algorithm is used for all individual clusters to analyze the distribution constructs of populations, which determines whether the new solutions come from global or local populations at a mating restriction probability. The mating restriction probability is updated at each generation. To adapt the balance between the population solution diversities and the convergence speed of the algorithm, at each generation, the control probability is adjusted by a developed adaptive strategy according to the reproduction utility of the two mechanisms of generating new solutions over the last certain generations. This research introduces the CoDE-MOPSO algorithm, designed to transcend existing multi-objective optimization methods’ limitations by optimally balancing exploration and exploitation. Our approach significantly advances evolutionary multi-objective optimization, demonstrating superior performance through lower Inverse Generational Distance and higher Hypervolume metrics, indicating enhanced efficiency in solving complex MOPs across various fields. In practical scenarios like gear reducer optimization, CoDE-MOPSO showcases remarkable effectiveness, highlighting its value in engineering applications and setting a foundation for sophisticated optimization strategies that combine speed with solution quality.