流程型工廠在雙評估準則下之排程研究

Abstract

流程型工廠排程問題中，決策者為了達成提昇系統績效之目標，除了 要使產出時間儘量降低之外，同時必須使在製品庫儲水準儘量降低，在這 雙重考量因素下，本論文將影響上述目標的主要兩項評估準則：(1).所有 工作完成時間(Makespan)最小化以及(2).總流程時間(Total Flow Time) 最小化加總合併，使之成為一個新的評估準則。並針對過去文獻較少探討 而實際生產環境必須考慮的動態排程問題，提出一藉由凍結事件發生時間 的方式，將動態排程問題轉換成為一個靜態排程問題的解決方法。本論文 針對雙機流程型工廠靜態排程問題以及雙機與多機流程型工廠動態排程問 題經轉換後的靜態排程問題分別提出啟發式排程演算法，並個別建立整數 規劃模式，以驗證各啟發式排程演算法所得結果之正確性，及作為評估啟 發式排程演算法求解品質成效的基準。當解大型排程問題時，基於求解效 率之考量，本論文所提出之啟發式排程演算法，能迅速而有效地獲得較佳 之可行解，以協助管理者執行排程之作業。 To increase a flowshop performance, to lower both throughput time and work inprocess (WIP) as much as possible is necessary. In general, the scheduling criterion on makespan minimization can effectively shorten the throughput time while the scheduling criterion on total flow time minimization can effectively reduce the WIP. This study attempts to minimize the weighted sum of the abovetwo scheduling criteria in a flowshop environment. Three sub-problems : (1)static scheduling with bicriteria problem in a two-machine flowshop, (2)dynamic scheduling with bicriteria problem in a two-machine flowshop, (3)dynamic scheduling with bicriteria problem in a multi-machine flowshop are tackled in this study. A frozen-event procedure is proposed to transformdynamic scheduling problems into static ones. For each sub- problem, the corresponding heuristic scheduling algorithm is developed. Integer programmingmodels are formulated to validate the performance of those heuristic schedulingalgorithms. Results demonstrate that the proposed algorithms can efficiently solve the flowshop scheduling with bicriteria problems.