查詢結果分析
相關文獻
- 應用螞蟻分群模式進行零件家族形成
- 物料需求計畫程式應用於家具製造之研究(2)
- 疏伐木製造家具之生產製程系統規劃研究
- 考慮單元負荷與機器途程的製造單元分群研究
- A Heuristic for Solving the Improper Assignment of Shared Machines in Group Technology Cell Formation
- 沖壓件之骨架表達法及其自動分類之研究
- 群組技術應用於物流中心之研究
- 群組技術於晶圓廠佈置之應用
- Using Evolutionary Computation Approach to Improve the Performance of the Fuzzy-ART for Grouping Parts
- 兩階段之群組佈置方法
頁籤選單縮合
題名 | 應用螞蟻分群模式進行零件家族形成=An Ant-Based Clustering Algorithm for Part Family Formation |
---|---|
作者姓名(中文) | 高有成; 李岳倫; | 書刊名 | 管理科學研究 |
卷期 | 特刊 民94.12 |
頁次 | 頁87-98 |
分類號 | 494.5 |
關鍵詞 | 群組技術; 零件家族形成; 螞蟻演算法; 群體智慧; Group technology; Part family formation; Ant algorithms; Swarm intelligence; |
語文 | 中文(Chinese) |
中文摘要 | 針對少量多樣化的生產模式,單元製造是最有效率的生產系統。而群組技術是單元製造系統常用的設計方法,其需要一個有效的零件分群方法,來進行製造單元初步設計。集群分析法為常用的零件分群方法之一,搭配特殊的相似係數,能集群相似度高的零件成為零件家族。集群分析法可分為階層式分群與非階層式分群兩種,但階層分群法容易產生鏈結效應,而非階層分群事先需給定零件家族數目。故在本研究中提出一個以人工螞蟻分群模式(AntClust)為基之零件分群演算法,來解決上述問題。本演算法利用螞蟻演算法的特性:群體性與隨機性,使零件分群結果不會在分群過程中過早決定,降低資料本身特性之影響。另外,本演算法不必預先給定群數,可直接利用螞蟻自我組織能力形成自然群組。本演算法已經被開發成軟體系街並利 用零件分群績效指標,對六個文獻案例進行測試,都得到相當好的單元形成績效。 |
英文摘要 | Cellular manufacturing is one of the major applications of group technology. It requires an effective part clustering approach for preliminary manufacturing cell design. One of famous approaches is the cluster analysis method, which uses similarity coefficients and clustering methods to group similarity parts into part families. Clustering methods are divided into two categories: hierarchical and nonhierarchical methods. Hierarchical methods are divided into two categories: hierarchical and nonhierarchical methods. Hierarchical methods often suffer from chaining effects, while nonhierarchical methods need a predetermined cluster number. The research proposes a part clustering algorithm that is based on an artificial ant clustering model (AntClust). The algorithm utilizes the characteristics of ants, congregation and randomness, to prevent grouping results from being fixed during clustering process and to reduce the effects of noisy data. Besides, the algorithm doesn’t need a predetermined cluster number. The algorithm has the ability of self-organization to form part families naturally. The algorithm has the ability of self-organization to form part families naturally. The algorithm has been developed into a software system. Six literature problems were selected to test the proposed algorithm with respect to groping efficacy. We fond that the algorithm is able to obtain better machine cell configurations than other approaches. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。