多目標製造設施佈置模式

Abstract

摘 要 設施佈置問題方案的評估與選取，乃是製造管理上重要的課題。本文旨在 多目標 的環境中，構建一個有效的佈置模式供製造管理決策上的參考。 本文所發展之多目標設施佈置模式，除了有效消除因子間量尺差異與計量 單位的不同外；並提供決策者一個設施佈置方案的優勢指標 (DI)，以確 保所得佈置方案解的品質。另一方面，藉由設施佈置方案優勢機率 ( ) 的提供，讓決策者對所得的佈置方案優劣性有一定的衡量標準。此外，決 策者對不同規劃目標所給予的權重，也勢必將影響最後佈置方案的選擇。 因此，本文利用折半與加倍法 (halving and doubling procedure) 構建 一個成對比較矩陣，並藉由一個有效的驗證程序來檢驗此成對比較矩陣的 一致性，以方便目標權重的計算。本文中所使用的檢驗程序是屬「事前檢 驗法 (prior test)」，有別以往學者所提之「事後檢驗法 (posterior test)」。最後，藉由模式的驗證與範例的描述，證明本文所提出的方法 確實可以有效地得到良好品質的佈置方案，提供決策者參考與選擇。 關鍵詞：多目標設施佈置問題、二次指派問題、優勢指標、優勢機率、折 半與加倍法 、一致性。 ABSTRACT Among the various aspects of manufacturing management, the facility layout problem is one of the most important issue. In this thesis, an efficient approach for the multi-objective manufacturing facility layout problem is proposed. In this thesis, a new multi-objective heuristic algorithm for resolving the facility layout problem is presented. It incorporates qualitative and quantit-ative objectives and resolves the problem of inconsistent scales and differenmeasurement units. An efficient index, the dominant index (DI), is developed and incorporated in this heuristic algorithm to guarantee the quality of pro- posed solutions. Moreover, a new measure of solution quality, dominant probab- ility (Dp), is offered to determine the probability of one layout is better than the others. Furthermore, this thesis also presents a design process for determinati on the objective weights. This process uses the halving and doubling procedureto construct a pairwise comparison matrix. A test procedure is proposed to identify the inconsistency of the pairwise comparison matrix. The proposed method belongs to 'prior test', which is different to the 'posterior test' proposed by Saaty. Finally, examples are illustrated individually to highlightthe proposed approachof the multi-objective manufacturing facility layout problem. Keywords: Multi-objective facility layout problem, quadratic assignment probl- ems, dominant index, dominating probability, halving and doubling procedure, and consistency.