The fresh food e-commerce cold chain logistics site selection problem is vital for optimizing distribution by identifying ideal warehouse locations and redeployment strategies to enhance efficiency and product freshness. This paper categorizes fresh fruit into domestic and imported types, aiming to minimize logistics costs while meeting timeliness constraints through integer programming. The solution involves improved K-Medoids to determine warehouse locations and Simulated Annealing to allocate city shops, fruit types, and quantities to each warehouse, creating an effective logistics transfer program.