Impact of business service modes on distribution systems: A reinforcement learning approach

Abstract

In this paper we identify several typical business service modes in a general distribution system composed of a single distributor and multiple retailers. For different service models, the business process can be different and each member of the distribution system may observe different level of information. Each member makes decisions to minimize its long-run-average cost or maximize its profit. A reinforcement learning algorithm is applied to obtain the decision policies and system costs. By comparing the system costs in various business service modes, we show numerically the impact of different business service modes on the distribution system. The results provide managerial insights into actual decision making in distribution systems. © 2007 IEEE.