| Summary: | 碩士 === 國立臺灣大學 === 工業工程學研究所 === 90 === As a part of the supply chain management system, an important function of the Demand Planning module is to provide a historical demand database, with which users can save or query their multi-dimensional demand data. It also provides the capability to perform aggregate or disaggregate forecasting. Based on the results of demand plans, users can proceed to production planning activities, such as safety stock level determination or capacity preparation.
In this research, we first introduce the concept of the demand planning hierarchy. Then, we describe and define different types of demand planning views. A symbolic representation system to express these views and the demand planning hierarchies (including demand planning path and demand planning tree) is also proposed. Based on the degree of demand fluctuation or the forecast accuracy, the evaluation metrics are defined to evaluate the performance of demand planning hierarchy. Finally, we formulate heuristic greedy algorithms to optimize the demand planning hierarchy performance, and an optimization method based on the dynamic programming approach is also developed. A demand data set from a semiconductor manufacturing company is used to test the algorithms proposed in this research.
|
|---|
