The Effect of VMI Efficiencies on Demand Forecasting－Reinforcement Learning Multi-agent Analysis for Supply Chain Model

• Main Authors: Ching-Hsiang Tseng, 曾敬翔
• Other Authors: Kune-Muh Tsai
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/17627217397835540959

碩士 === 國立高雄第一科技大學 === 運輸倉儲營運所 === 91 === With the development of information technology, varieties of products and variation of customer demand make organization not have sufficient information to respond customer’s demand. Furthermore, uncertainty about short-run product demand also build up obstacles to forecast. Therefore, it is a hot issue to set up a model which can predict more accurate downward-stream demand. A simulation model embedded five demand forecast models: Moving Average Forecasting Model, Time Series Forecasting Model, Constant-value Forecasting Model, Stage-Learning Genetic Algorithms Forecasting Model, Continue-Learning Genetic Algorithms Forecasting Model, is used to simulate and analyze the proposed VMI supply chain system. Among the five demand forecast models, the prediction error of Continue-Learning Genetic algorithms Forecasting Model is smallest. Therefore, variation of inventory level in Continue-Learning Genetic algorithms Forecasting Model is the most stable also.