Optimization and Simulation of Distribution System in a Supply Chain Based on Q-learning

A distribution system consisting of multiple manufactures and one retailer was studied in this paper. In this system, the demand of the manufacturers´ products is sensitive to service level, and their products have substitution. The manufacturers motivate the retailer to maximize their profits by providing bonus. The retailer provides corresponding service level to maximize its profits by obtaining bonus from manufacturers. A simulation was done to this system under both non-cooperation and cooperation situations based on Q-learning algorithm. The simulated results show the manufacturers´ profits are damaged in non-cooperation situation. The manufacturers would be benefited as they adopt the cooperation action, but the retailer´s profit is damaged