<p>In this study, a collaborative optimization framework combining artificial fish swarm algorithm (AFSA) and Voronoi cell finite element method (VCFEM) is proposed to solve the problem of the influence of microstructure distribution on the macroscopic mechanical properties of particle-reinforced composites. Compared with the traditional displacement-based finite element method, VCFEM significantly improves the computational efficiency of multi-inclusion problems by means of generalized stress function and element local adaptation technology, and accurately captures the stress field discontinuity and interface stress concentration phenomenon. At the same time, AFSA abandons the dependence of traditional optimization methods on simplified models, and directly searches for the global optimal solution through discrete topological variables (particle positions). Its natural selection mechanism and swarm intelligence characteristics effectively overcome the local optimal trap. The numerical simulation results show that the proposed method exhibits high accuracy in both single- and multi-inclusion models (the error with the commercial software MARC is less than 5%), and in the complex model with 100 inclusions, the maximum Mises stress is reduced by 32.6% after optimization. The synergistic effect of the two breaks through the trade-off between efficiency and accuracy in traditional optimization, and provides the ability of both computational efficiency and global optimization for multi-scale composite material design.</p>