透過您的圖書館登入
IP:18.117.137.64
  • 學位論文

使用蟻群系統求解協同式製造資源規劃之研究

Resource Planning for Collaborative Manufacturing Using Ant Colony System

指導教授 : 鄭宗明
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


現今的生產模式,面對產品訂單快速的異動,已逐漸趨向於小批量多樣化的訂貨生產(Make-To-Order)方式,其優點為產品種類多,因此兼顧生產現場資源的最佳化成為現今面臨的最大挑戰。而傳統製造資源規劃在面臨複雜的生產環境,如果製造廠無法快速的獲得資訊、迅速的處理資訊及有效的產生規劃,則會在激烈的市場競爭中沒有生存空間。 在全球化的製造行為上,為了減少成本及時間的浪費,必須協調整個供應鏈的管理,而在管理上,可藉由協同的行為模式來管理供應鏈中的企業,主導廠和協力廠之間的關係,乃由主導廠進行生產規劃時,在將生產資訊主動提供給協力廠,其生產資訊包括種類、數量、規格、交期等,此一關係即為協同式的關係,透過此一關係,主導廠藉由連結各個不同的協力廠而形成龐大的虛擬企業(Virtual Enterprise, VE)。 為了達成上述虛擬企業之架構,本論文採用了一種模仿自然界中螞蟻覓食機制而發展出的啟發式方法稱為蟻群系統(Ant Colony System, ACS),利用人工螞蟻所創造出來的群體智慧(Swarm Intelligence)幫助虛擬企業選擇協同夥伴並指派任務,使排出可行之製造資源分派方式或是訂單分佈方式,令生產成本降低。本論文利用該演算法中分散處理與快速累計的優勢,應用在協同製造資源規劃(Collaborative Manufacturing Resource Planning, CMRP-II)之問題上,於求出任務指派方式後,找出成本最低之前三名,再利用模擬軟體eM-plant建構生產系統並模擬之,以驗證螞蟻演算法之生產規劃。完成基本規劃後,本研究亦針對內部及外部異動因素訂定了局部重新規劃之機制,最後再供作決策者執行試做和決策之參考。

並列摘要


The market today demands wide variety with quick turnover, making Make-To-Order a suitable production policy. The objective is to optimize the resource allocations using the realtime data that market competitions based on. A virtual enterprise always makes use of the production resources provided by its collaborators to generate maximal profit. The organization of the collaborations is managed by balancing the production needs with the available resources. As a result, the type and amount of tasks dispatched, the production time acquired, and the deliver time can be quantitatively estimated for decision references. In this research, a heuristic Ant Colony System method is implemented on a virtual enterprise to organize the collaborative resources and to ensure minimum cost. During the process, the algorithm will generate the top three candidates for further evaluations using a simulation tool, and then determine the optimal solution. The method also provides thresholds to active automatic rearrangements when the conditions within the production system and among the collaborations have changed. In the implementation, MS-ACCESS is used to create the collaborative database, Matlab is used to create and perform the artificial swarm intelligence, and eM-Plant is used to do the simulations. The results showed that the method can generate instant decisions with outstanding performance.

參考文獻


林依潔,「整合模糊理論與螞蟻演算法於含時間窗限制之車輛途程問題」,國立台北科技大學,台北(2003)
盧銘勳、郭政宜,「應用螞蟻演算法於動態排程系統」,臺北科技大學學報第四十一之二期,逢甲大學,台中(2008)。
Alvarez, E., An application of a real-time scheduling system for turbulent manufacturing environments, Robotics And Computer-Integrated Manufacturing, Vol. 20, pp. 485-494, 2004.
Chen, J.H. and Ho, S.Y., A novel approach to production planning of flexible manufacturing systems using an efficient multi-objective genetic algorithm, International Journal of Machine Tools & Manufacture, Vol. 45, pp. 949-957,2005.
C. Blum, Ant colony optimization: Introduction and recent trends, Physics of Life Review2, 353-373, 2005.

延伸閱讀