The current study puts forward the idea of decision-making based on collaboration between the demand forecasting and logistics coordination process of fresh food, using improved PSO-BP neural network. The BP neural network is reinforced through application of PSO technology that incorporates optimized inertia weight, environmental awareness, and solution jumping technique, thus enhancing the accuracy of fresh food demand prediction. Also, the study formulates the multi-constraint and multi-objective optimization model with consideration of time and cost-related factors. The proposed model is then solved using simulated degradation ant colony algorithm, thereby widening the space for finding the optimal solution. The empirical analysis results reveal that the IPSO-BP model demonstrates better performances in comparison with ordinary BP and PSO-BP models in terms of various criteria. For instance, applying the multi-objective distribution model in City A leads to emergence of five optimization routes and decreases the logistics costs, complying with timeliness and load capacity requirements. Thus, the study makes important theoretical and practical contributions to the development of intelligent collaborative optimization of fresh food logistics supply chain.
