訂單式生產系統中動態控制訂料時機之研究

Optimal Control of Ordering Epoches for a Make-to-Order Inventory-Production System with Random Lead Times for Raw Material Orders

洪瑛香奉(Ying-Feng Hung)

指導教授 ： 張國華

中原大學/工學院/工業工程研究所/碩士(2003年)

https://doi.org/10.6840/cycu200300259

摘要

現行的生產管理系統，可分為計劃式生產(Make-to-Stock, MTS)、訂單式生產
(Make-to-Order, MTO)。由於網路的發達與電話服務中心的興起改變了消費者對商品
的需求型態，再加上企業產品相互競爭，產品需求趨於多樣少量、生命週期日短。
所以富彈性化、低成本、反應快等將為未來生產系統的重要特性。近年來隨著產業
環境型態的改變，再加上電腦大廠戴爾(Dell)的影響，於是許多產業主動或被動地將
部份生產或下單方式由計劃式改為訂單式。在一訂單式生產系統，由於訂單的不確
定性，對於原物料庫存的掌握不易：過少的庫存可能造成延遲交貨或訂單流失的損
失；過多的庫存造成資金的積壓又可能因產品週期短而會形成呆料。面對訂單的不
確定性，傳統的方式如訂單需求預測與靜態的管理將逐漸轉為動態地管理庫存與備
料。為了避免可能會造成上述之損失，如何精準地設定原物料的訂購點，是個非常
重要的議題。面對日亦複雜以及反應快速的生產作業系統，若只依據個人單獨經驗
與直覺無法完善地管理。我們需要藉由數量的方式輔助管理決策的訂定。
本論文將以動態的方式控制訂單式生產的庫存，我們將參考原物料庫存與訂單
數決定訂貨時間點以使總成本為最小。本論文將以量化的方式決定庫存與物料管理
中訂貨的方式。我們以等候系統描述訂單式生產系統。在考量兩個不確定性的因素
(顧客的需求以及訂購原物料的前置時間)，我們以馬可夫決策過程(Markov Decision
Process, MDP)依等候的訂單數目與現有的存貨量下找出最合適的訂購時機與存貨
水準，進而分析此最佳訂購時機策略的結構，例如Switching Curve 之性質。我們也
將撰寫數值求解程式應用於實際問題。

關鍵字

訂單式生產系統 ； 馬可夫決策過程 ； 最佳策略 ； 庫存管制 ； 等候線系統

並列摘要

Production system can be distinguished as a Make-to-Stock (MTS) system or a
Make-to-Order(MTO) system. Due to the fast-developing computer networks and
the raising of the call centers, a customer can put his custom order through the phone
or Internet. The need for specialized, custom products has been growing as a result
of shorter product lift cycles and new manufacturing technology. MTO
manufacturing has become an important sector of the industry.
In a MTO system, because of the uncertainty of the demands, it is important to
control the stock of raw materials. Due to the uncertainty of the demands, the
traditional approaches such as the forecasting of product and the static inventory
control are replaced by the dynamic control approaches. Therefore, how to decide
the reorder points for the raw materials becomes a very important issue.
This thesis studies an inventory replenishment policy, in which demands arrive
to the production workshop according to a Poisson process, and are processed in an
FCFS manner, and only one unit of raw materials is ordered at the reorder points.
Based on the inventory level of materials and the accepted MTO orders, we
dynamically select the reorder points in order to minimize the total discounted cost
of the production system. We model our optimization problem as a Markov decision
process, we prove the monotonic properties of the optimal policy and study the
property of the underlying switching curves. We also develop numerical procedures
for solving the real problem.

並列關鍵字

Make-to-order production system ； Markov decision process ； Queueing system ； Optimal policy ； Inventory control

參考文獻

[6] K.-H. Chang and T.-L. Lee, Optimal Admission Control to Two Multiserver

[1] S. Axsater, Continuous Review Policies for Multi-level Inventory Systems with Stochastic Demand, in: Logistics of Production and Inventory, Handbooks in

Operations Research and Management Science, 4, 195-198, 1993.

[2] O. Berman and K. P. Sapna, Optimal Control of Service for Facilities Holding Inventory, Computer and Operational Research, 128, 429-441, 2001.

[3] D. P. Bertsekas, Dynamic Programming, Deterministic and Stochastic Models,