- 學位論文
應用蟻群演算法求解多目標動態設施規劃問題
Apply Ant Colony Optimization to Multi-Objective Dynamic Facility Layout Problem
摘要
由於現今經濟的發達，製造業也隨之蓬勃發展。對於產品需求量不斷增加，製造業工廠之運作，為了提升生產效率以及節省大量成本，以獲得最大的利潤為目標而努力，因此設計良好的設施規劃及持續性的改善，成為本研究最重要之目的。而為能符合實際情況，本研究將多目標決策與動態設施規劃做結合，並考量部門之間總物料流動成本與鄰近關係重要度評比兩目標。發展出多目標動態設施規劃之問題，以考量更多的層面且更複雜的因素，使本研究能更貼近實務之應用。 多目標動態設施規劃此類二次分配問題，為一個NP-Hard問題，若以啟發式演算法計算求解將節省大量時間與成本，因此本研究運用混合式蟻群系統(Hybrid Ant colony System, HAS)，結合多目標決策-柏拉圖最佳前緣曲線(Pareto optimal front)，將其非支配解之概念導入蟻群最佳化演算法，發展出柏拉圖最佳化之多目標動態蟻群最佳化演算法(Ant Colony Optimization algorithm for Pareto optimal Multi-objective Dynamic Layout, ACO-PMDL)，以求得多目標動態設施規劃之最佳解。 利用本研究之ACO-PMDL，分別對於文獻中動態設施規劃問題、多目標設施規劃問題以及三組不同相關性的多目標動態設施規劃案例之數據，以測試ACO-PMDL求解之表現。在測試結果中顯示，ACO-PMDL對於上述前兩種問題測試求得之最佳解，相較文獻的最佳解有著較佳的品質。而對於多目標動態設施規劃問題，此ACO-PMDL所求得的解，能有效地將相鄰部門之間符合其關係重要度。在總物料流動成本表現上，更能優於文獻中的表現。由測試結果說明，本研究之柏拉圖多目標動態設施規劃之蟻群最佳化演算法，可有效搜尋出多目標動態設施規劃問題之最佳部門排列。
並列摘要
With today’s booming economy, in order to keep with product demand, manufacturing factory must find a way to raise productivity and reduce cost to obtain maximum profits. Thus, the purpose of the study is to improve facility layout. In order to reflect the reality situation, this study combines multi-objective and dynamic facility layout, considing the both material flow cost and closeness ranking between departments. Therefore, Multi-objective dynamic facility layout problem is developed. Multi-objective dynamic facility layout problem is a NP-Hard problem, which and resembles the quadratic assignment problem. It uses “Heuristic” to solve NP-Hard problem to save production time and cost. Therefore, this study uses hybrid ant colony system with Pareto optimal front. The Pareto optimal front is combined with Ant Colony Optimization to form Pareto optimal Multi-objective Dynamic Layout (ACO-PMDL) to solve the Multi-objective Dynamic Facility Layout problem. In this study, we use ACO-PMDL method to solve for dynamic facility layout problem, multi-objective facility layout problem and Multi-objective dynamic facility layout problem, based on datasets from literatures to test ACO-PMDL solution quality. The results show ACO-PMDL has better solutions for dynamic facility layout problem and multi-objective facility layout problem. For multi-objective dynamic facility layout problem, the ACO-PMDL solution effectively places one department next to another in terms of importance departments. In terms of total material flow cost, the results from ACO-PMDL are much better than reference solution. From the results, ACO-PMDL of this study is proven to find the best department assignment of Multi-objective dynamic facility layout problem.