Modeling and Simulating for a Distribution System Based on Fuzzy Systems to Forecast Demand

Abstract

A logistics cost model based on demand forecasting is proposed for a two-echelon distribution system with a central warehouse and multiple retailers in this paper. The retailers and the central warehouse all use periodic control policy to review their inventory level. First, we attempt to develop a fuzzy system-forecasting model capable of learning the IF-THEN rules obtained from demand data and experience of marketing experts with respect to promotions. Then we build a comprehensive model to combine demand forecasts with inventory decision and distribution system cost model. Finally, fuzzy system-forecasting model is compared to conventional regression method by a numerical example and its results indicate that the proposed fuzzy system-forecasting model performs more accurately than the conventional regression method. The computational results also show that substantial cost savings and improved service level can be realized through applying fuzzy systems to forecast demand.