考量機台損耗之連續工作站動態派工

Abstract

本研究考量在由非等效平行機台構成之串聯式生產系統中的派工問題。非等效平行機台代表機台間能進行相似的加工程序，但其加工速率並不相同，而在本研究中，該項差異來自於機台間健康狀態的不同。為最小化周期時間(Cycle time)，本研究結合線性規劃與動態規劃以發展混合式淨空動態派工(HCSDD)。我們將系統中的各工作站拆解成單一工作站，並以線性規劃評估其最大平均產能，再根據工作站位置及最大平均產能決定在單站動態派工中使用的參數及決策方法。透過對整體系統的拆解，求解複雜度被有效降低，使HCSDD在提供良好的整體派工決策的同時，保持其計算可行性。

This research studies scheduling problems in multi-stage production systems with unrelated parallel machines. Unrelated parallel machines are machines functioning similarly but with different processing rates, which are caused by machine deterioration. To minimize average cycle time, Hybrid Clearing System Dynamic Dispatching(HCSDD) was developed by combing linear programming decomposition with dynamic programming. The whole system is decomposed to single workstations and the capacity of each workstation is estimated by a linear programming model. For each workstation, the dispatching policy is obtained from single workstation dynamic dispatching model. The parameters used in the model and decision rules are determined based on the workstation location and capacity. The algorithm is computationally efficient and performs well in large systems. Comparing with other methods, cycle time is significantly improved in simulation analysis.