Multi-attribute Dispatching Rules to Improve Flexible Manufacturing System Performance

[[abstract]]廣為製造業所使用來因應產品多樣少量需求的彈性製造系統（Flexible Manufacturing System 簡稱FMS）在生產的流程中，加工時間佔的比例通常不大，相對花費在運送、等待的時間則佔有相當大的比例，而無人搬運車(Automated Guided Vehicle AGV)為其中一種常見之物料搬運設備。在系統中調度工件的派工法則和AGV的派車法則更是影響績效的關鍵，在過去文獻中，提出了許多調度法則，有單屬性也有多屬性的指派，然而至今仍沒有任何一套法則能滿足能夠達到所有績效最佳化的目標。所以本研究目的在於了解FMS中不同環境對法則指派的影響，以此來結合多屬性派工與多屬性的派車法則並提出混合的權重建議，希望能產生近似最佳解的表現，又同時滿足多重的績效指標。 本研究運用系統模擬(systems simulation)方法建構一彈性製造系統模型，並以實驗設計規劃模擬實驗參數，探討在不同環境下法則對績效的影響，並將設計的多屬性指派法則與一般常用的法則比較其優劣，實驗結果發現，在多目標指派問題中，法則考量的屬性越周全，績效將越優異。而僅考慮單屬性之派工與派車法則或許可使某特定績效指標達到最佳表現，但於它項績效指標之表現則多半不佳。若以整體績效而言，無論於何種環境，結合多屬性派工與派車法則在多個目標績效表現下是較為優異的。[[abstract]]Many manufacturers adopt multiple-type and small-quantity production systems such as Flexible Manufacturing System(FMS) In a manufacturing process the time actually spent on the machines is sometimes not very long; on the other hand a significant portion of the whole process time is consumed for transporting and waiting Automated Guided Vehicle(AGV) is common of Material Handling System equipment adopted by many companies Scheduling of the jobs in the system and the AGV dispatching rule is a key of affecting performance In literature there are many types of dispatching rules such as single- and multi-attribute dispatching rules However a dispatching rule that is good for all cases does not exist Therefore this research aims to understand the different environmental effects on the rule in order to combine multi-attribute dispatching rules with recommendations weights hoping to generate near optimal performance solution but at the same time meet multiple performance indicators In this research first constructed simulation models of Flexible Manufacturing System and applied Design of Experiments to perform simulation experiments to discover the effects of dispatching rules on the performances in different scenarios We propose a new multi-attribute dispatching rule for this type of environment and compare their performance with that the best dispatching rules in literature Using simulation we show that the new multi-attribute dispatching rules will perform well when considering more attributes In terms of overall performance whether in any environment combined with the multi-attribute dispatching rules performance well on multiple objective