設備動態配置與預防保養之整合研究

Abstract

本研究整合串連式生產系統（serial production systems）內的動態預防保養及動態多功能設備配置決策。由於生產系統多面臨外部的需求不確定，及系統內部的加工時間與設備可靠度不確定，因此如何整合預防保養與多功能設備配置有其必要性。本論文結合動態規劃（dynamic programming）與啟發式演算法（heuristic algorithm），發展有效的預防保養與動態配置決策工具，並利用離散事件模擬（discrete event simulation）進行驗證。 本論文主要可分為兩階段，第一階段利用動態規劃求解出小型兩階段系統的最佳動態資源配置方法，並以此為基礎發展啟發式動態設備配置與動態啟發式預防保養策略；第二階段將結果延伸至大型四階段串連式系統，並將結果與過去文獻之動態配置方法相較，系統產出(throughput)提升約10%，平均等候成本(average holding cost)大幅降低約20%，在製品在系統平均時間(average cycle time)也降低約22%，發現結合動態配置和預防維修可以更有效率的使用設備提生產出，並且改善系統阻塞的情況。

This paper combines dynamic resource allocation and preventive maintenance to solve “Demand uncertainties” and “Resource reliability” in a serial production system. Because of demand uncertainties and resource reliability, when and how to do preventive maintenance and allocate resource is very important. In this paper, we use dynamic programming and heuristic algorithm to develop an effective preventive maintenance and dynamic resource allocation strategy. This paper can be divided into two stages. In the first stage, we use dynamic programming to solve a Markov decision Process in a two-stage system. Then we use the result of two-stage system to develop dynamic heuristic resource allocation and dynamic heuristic preventive maintenance method. In the second stage, we apply the result to more complex system and compare the result with other allocation method. We find that throughput is increasing around 10%. Average holding cost and average cycle time are decreasing over 20%. To sum up, the combination of dynamic resource allocation and dynamic preventive maintenance can use resource in a more effective way and also decrease bottlenecks in a system.