When dealing with large-scale logistics scheduling problems, the computational complexity of traditional methods increases sharply, resulting in an excessively long solution time and making it difficult to meet the real-time requirements. In order to overcome the limitations of traditional logistics scheduling methods, intelligent optimization algorithms emerged, and PSO is a typical representative among them. However, PSO also has some deficiencies in practical applications. This research is based on the PSO algorithm and combined with the actual needs of logistics scheduling. During the particle update process, the 2-opt algorithm is introduced. By exchanging two edges in the path to eliminate the intersection, the total path length is reduced. Combined with threshold acceptance, random restart or other heuristics, the global search ability is improved. In order to verify the improvement effect, two baseline models, PSO and GA, were selected for comparative experiments, and a comparative analysis was conducted on the key indicators of logistics scheduling in Logistics Enterprise A. The experimental results show that IPSO has advantages in terms of delivery time, cost and vehicle utilization rate.