產能配置與設備維修之動態整合研究

Abstract

動態產能配置與設備預防保養排程規劃中存在幾項不確定因素，包含：外部需求的不確定性與資源設備本身之可靠度問題等，皆影響企業運作之產出與成本，因此本研究將針對整合產能配置與設備維修所面臨的兩項不確定因素加以探討，目的為解決各規劃期間內，不同系統狀態下之資源配置與維修策略，以達企業成本之最小化。本研究與過去文獻相異之處在於資源配置與預防維修保養兩類決策為同時進行成本考量與執行。 本研究考慮多功能資源設備可處理系統中多階段製程之特性，利用動態規劃方法，針對多功能資源動態配置之數學模式，輔以數值迭代法（Value Iteration Algorithm）進行模式求解，制定產能配置規劃期限內各種狀態之最適配置策略，以達成最小化等後成本為目標。文中並發展一個啟發式成本比較利益方法（Heuristic of Weighted Cost Comparison method），用以將求解策略應用擴大至多階段串連式等候線系統，藉由eM-plant模擬軟體進行不同方法之間的驗證與比較分析，同時再進一步探討整何預防維修保養方法，使系統運作更有效率與大幅降低等候成本。

Our Research consider the coordinated preventive maintenance and allocation of reconfigurable resources in a serial line problem. There are two kinds of uncertainties factors affect the decision planning. These factors include external demand uncertainty and internal machine reliability. The objective of the research is decide the srtategy of resource allocation and preventive maintenance action at the same time to minimize the average holding costs. We model the dynamic resource allocation problem with Markov decision process (MDP). In this MDP model, we consider a two stage tandem queue system with machine failure and solve it by value iteration algorithm. The solution provide flexible servers allocation strategies to achieve minimize holding cost optimality. In this paper, we develop a heuristic of weighted cost comparison method (HWCC) to connect the two stage policy and expand a four stage system. HWCC is compared to several heuristics from the literature by eM-plant simulation software. We also used HWCC to integrate preventive maintenance policy that improve the performance significantly.