The work aims to optimize the cold chain distribution path of fresh food under the regional time-varying road network, to reduce the distribution cost of enterprises and improve the distribution efficiency. According to the timeliness and temperature management requirements of cold chain distribution, the regional time-varying vehicle travel function combined with road congestion was constructed, and the optimization model based on cold chain distribution cost of fresh food was established. The improved particle swarm optimization algorithm (PSO-GA) was used to solve the problem, and the optimization results of regional time-varying model, time-varying model and static model were compared. PSO-GA was superior to particle swarm optimization (PSO) and genetic algorithm (GA) in both accuracy and efficiency. In terms of total distribution cost, PSO-GA was 2.0% lower than PSO and 4.2% lower than GA. In terms of carbon emissions, PSO-GA was 3.9% lower than PSO and 11.2% lower than GA. The model can successfully reduce the distribution cost and carbon emission cost in the complex regional congestion environment, and can better simulate the reality of road traffic distribution, which has a good practical significance.