全自動化生產單元 ( Fully Automatic Manufacturing Cells， Crama et al 2000 ) 由自動化加工設備及自動化搬運系統(機器手臂、天車….)所構成，已經成為製造系統中常見的生產設備，生產單元內自動搬運系統的效率與整個系統的效率息息相關，而加工設備與搬運設備間密不可分的關係及其複雜的交互作用更導致除了硬體設計外，自動化程式在設計上的困難性。 在眾多考慮到自動搬運系統的問題 ( 包括Robotic Cell Scheduling 、Scheduling of AGV ) 中，天車排程問題 ( Hoist Scheduling Problem-HSP， Phillips and Unger 1976) 或吊車排程問題( Crane Scheduling Problem， Lieberman and Turksen 1981 ) 是受限最多的問題 ( Crama 1997 ) 。這個問題已經由Lei證明其NP hardness ( Lei and Wang 1989 ) 。尤其在絕大多數實際上運用的系統均是採用〝非正式〞的方法，依經驗以嘗試錯誤法來獲得解決。通常需要花上數日的嘗試並配合實際硬體加以測試，才能獲得可接受的結果。 目前多數對於自動化表面處理線『天車排程問題』的研究均集中於在單一天車、單一產品的環境下，以事前規劃的方式，求得產能最佳化的天車行程。但實際系統通常運用多部天車同時運作，而且這種針對單一產品所規劃的行程會遭遇到以下三種主要的問題：(1)單一生產線只能生產單一規格的產品; (2)在某些無法完全避免的機械故障的情況下(如某一天車或某處理槽故障)，形成原規劃天車行程無法應變，而造成大量產品報廢的情形。(洪瑞炫 2002) (3) 先前的研究(Mak et al 2002)指出多部天車間部份搬運區的重疊(overlapping)可能可以提昇天車的搬運效率，但未有進一步的結論。故此研究的目的在討論能彈性反應系統動態，在上述情境發生的狀況下，能作為靜態天車行程規劃外可供切換的「天車指派法則」。 至於控制結果的分析方式，對於單一天車系統，Phillips and Unger (1976)，Shapiro and Nuttle (1988) 等提出之實例已經成為過去許多研究的標準測試例題(Benchmarks)，但對於多部天車的研究或實例卻相當少，本研究所討論的即時控制方法，將以Mak et al.(2002)所提出的案例加以研究分析。 There are a lot of researches which consider scheduling and control of automated material handling systems, including robotic cell scheduling、scheduling of automated guided vehicle and so on. Among these problems, the hoist scheduling problem-HSP or crane scheduling problem is the most restricted one (Crama 1999). And it was proved to be NP-hardness in the class of this problem by Lei and Wang (1989). Most of the systems installed are designed by experienced experts using trail-and-error based techniques. The fixed schedule derived by trial-and error must be validated with an actual system and usually it takes several days to correct some unexpected errors in the original design. The previous study about the HSP in automated surface treatment lines mainly focus on single hoist、single product type environment. A fixed hoist schedule for maximum through-put was considered before the operation. The difference between previous case studied and actual systems is: real systems are usually composed by several hoists, and it will encounter three difficulties with a fixed schedule design: (1)the system could only operate with one product type at one time (2)it will not work if one of the hoist fails. Therefore there will be lots of scrap products if only one hoists fails(3)a previous study (Mak 2002) suggest that adding overlapping zones between hoists movement could be beneficial to material handling efficiency but no further conclusion. The purpose of this study is to develop an efficient hoist dispatching rule that deal with above scenarios in a automated electroplating line. To analyze the performance of different approaches, some of the single hoist cases are discussed extensively by many researchers. For example, the cases provided by Phillips and Unger(1976)、Shapiro and Nuttle(1988). And these cases already become benchmark problems in this field. But it is rare to study a multi-hoist system. We will use the multi-hoist case provided by Mak(2002) to study the performance of our rule.
