Solving the MDVRPTW problem not only requires optimizing vehicle routes but also must satisfy the time window constraints between multiple depots and customers. To address the limitations of traditional heuristic algorithms when solving the multi-constraint multi-depot MDVRPTW problem, this paper proposes the IHTS algorithm by combining Particle Swarm Optimization (PSO) and Tabu Search (TS). The algorithm significantly improves performance through optimized encoding, improved neighborhood search, and the introduction of a two-layer tabu table. Experimental results demonstrate that the IHTS algorithm outperforms traditional methods in several test cases, providing a better solution to the MDVRPTW problem with strong practical application value.