臺灣博碩士論文加值系統

在當前的社會中，顧客對於產品的要求，不再滿足於充足而價廉的產品，轉而要求精緻、與眾不同的多樣化產品。換句話說，大量生產已經不再完全適用於今日的製造業。「小批量、多樣化」的生產形態(批量生產)已是目前生產的主流，但是由此轉變所引發的問題卻不少，例如：成本的提高、資源的浪費、在製品存貨的增加、工作排程的困難以及物流的煩瑣….等等。這些都顯示出批量生產形態需要一套良好的生產系統或技術互相配合以降低成本及提高生產力。而群組技術將相似設計或製程工件的合併製造以改善其設計、生產效率，並提高資源使用，擷取相似製程的利益。因此實行群組技術可在少量多樣的生產型態下達到大量生產的經濟效益。
為達成群組技術的目的，本研究提出了結合粒化運算、遺傳演算法和限制滿足神經網路觀念於群組技術中群化模式建立。運用此方法能處理二元值的機器/工件矩陣問題，而且問題本身能夠以多途程間的競爭性及粒化運算本身的關聯性來分群。簡單的來說，就是產生一個適當的細粒數目來涵蓋所有的資料以及在限制滿足一個工件只使用一個途程下使得途程最佳。也就是說，本研究方法藉由所提之演化限制滿足模式(Revolutionary Constraint Satisfaction model, RCS)，拓展了處理傳統單元形成問題的能力，而具有處理多途程之問題。本研究方法是二階段模式。第一階段主要以一個工件只有一個途程之觀念來形成途程集合。而第二階段是以第一階段所形成之途程集合來將工件分族及機器分群。
本研究所提出之群化模式具有許多優點，特別是此方法減少了複雜的計算和具有處理大型分群問題的能力，同時可降低落入區域極值的機率，求得近似最佳解。而本研究的研究貢獻主要在於有效結合遺傳演算法與粒化運算、類神經網路之概念處理來處理大型的單元製造問題。本研究的創新觀念包含融合類神經網路能量函數於遺傳演算法中，以求得最佳途程問題；以及結合粒化運算概念於遺傳演算法，以求得最佳分群問題。本研究方法利用數個有名的範例驗證其精確度。

In the recent world, the requirements to product from the customers are no longer the satisfaction of sufficiency and cheapness, but the provision of variety of delicate and extraordinary products. In other words, mass production of a uniform product is not so appreciated by the customers. Thus, the new principle of “lot size production” has become the main stream of manufacturing philosophy. Nevertheless, the problems sprouted by this change are significant so as to the increase of costs, the waste of resources, the increase of work-in-process, the difficulty of scheduling and the complexity of material handling. All of the problems caused by lot size production indicate the necessity of a sophisticated production system or technique to reduce the costs and increase the productivity. Group technology, which combines the similar parts or machines in design or manufacturing into families or machine cells, can improve the efficiency of design and production because of the full utilization of resources and the advantage of similar manufacturing processes.
To complete the purposes of group technology as mentioned, this research attempts to propose a clustering analysis model, which integrates granular computing method, genetic algorithm, and the concept of constraint satisfaction neural network. In this model, the problem is simply considered as the clustering analysis for a two-dimensional data, an incidence part/machine matrix with multiple process plans for each part. The problem itself is inherently association with machine granules and machine granulation with the involvement of the competition of multiple process plans. Simply speaking, one may think it is to produce an appropriate number of granules that cover almost all of the data and to maximize the goodness fit of designing the process plans for the parts under the constraint satisfaction of each part with only one process plan. 　In other words, this research method expends its capability of handling the cellular formation problem with the involvement of multiple process plans through a proposed Revolutionary Constraint Satisfaction model (so-called RCS model). It is a two-stage model. The first stage is to organize a set of process plans in which each part must and only has one process plan. The second stage is to cluster the parts into families and the machines into cells based on the set of process plans selected in the first stage.
The advantages of this technique are the simplicity of computation, the capability of handling a large scale of clustering analysis, and the reduction of the possibility of falling local minimum in the solution searching process. The main contribution of this research is the effectiveness of integrating genetic algorithm, granular computing and the concept of neural network for dealing with large-sized cellular formation problem. The novel ideas include embedding the energy function of neural network in the genetic algorithm for multi-process plans problem and combine the granular computing concept with the genetic algorithm for cellular formation. This proposed model has been verified and confirmed by its accuracy using several popular cases.

誌謝 I
摘要 II
ABSTRACT III
目錄 IV
表目錄 VIII
圖目錄 XI
第一章 緒論 1
1.1 研究動機 1
1.2 研究目的 4
1.3 研究流程 5
1.4 研究範圍與限制 6
1.5 章節概要 7
第二章 文獻探討 8
2.1 群組技術(GROUP TECHNOLOGY, GT) 8
2.2單元形成(CELLULAR FORMATION) 9
2.2.1檢視法與分類系統法 9
2.2.2生產流程分析法 10
2.3求解最佳化問題的探討 18
2.3.1區域搜尋法 19
2.3.2典型搜尋法 19
2.3.3區域搜尋法與典型搜尋法的比較 20
2.3.4區域搜尋法的搜尋型態 21
2.4遺傳演算法(GENETIC ALGORITHM, GA) 22
2.4.1基本原理與起源 22
2.4.2遺傳演算法的演算流程 23
2.5粒化運算(GRANULAR COMPUTING) 26
第三章 研究方法 27
3.1研究架構 27
3.2問題模式 28
3.2.1方法說明 29
3.2.2符號說明 29
3.3演算方法 31
3.4演算流程 32
3.4.1演算流程(一)：演化限制滿足模式 34
3.4.2演算流程(二)：粒化遺傳演算模式 41
3.5演算實例 47
3.5.1實例前半部(演化限制滿足模式) 47
3.5.2實例後半部(粒化遺傳演算模式) 57
第四章 範例與演算結果分析 64
4.1參數設定 64
4.2範例資料說明 65
4.2.1範例(1) 66
4.2.2範例(2) 73
第五章 雛形系統說明 82
5.1系統概論 82
5.2系統實例說明 82
第六章 結論與建議 86
6.1結論 86
6.2研究貢獻 87
6.3對後續研究建議 87
參考文獻 88
作 者 簡 介 92

英文部份：
〔1〕Academic, K., Group Technology and Cellular Manufacturing : A State-of-the-Art Synthesis of Research and Practice, Boston, PP.278-279(1998).
〔2〕Adil, G. K., Rajamani, D., and Strong, D., “AAA: An Assignment Allocation Algorithm for Cell Formation,” Working Paper, University of Manitoba, Canada(1993).
〔3〕Aljaber, N., Baek, W., and Chen, C. L., “A Tabu Search Approach to the Cell Formation Problem,” Computers and Industrial Engineering, Vol.32, No.1, PP.169-185(1997).
〔4〕Chandrasekharan, M. P., and Rajagopalan, R., “MOD-ROC: An Extension of Rank Order Clustering for Group Technology,” International Journal of Production Research, Vol.24, No.5, PP.1221-1233(1986).
〔5〕Glover, F., “Future Paths for Integer Programming and Links to Artificial Intelligence,” Computers and Operations Research, Vol.13, No.5, PP.533-549(1986).
〔6〕Goldberg, D., Genetic Algorithms in Search, Optimization, and Machine Learning, world scientific, PP.125-127 (1989).
〔7〕Gupta, T., “Clustering Algorithms for the Design of a Cellular Manufacturing
System: an Analysis of their Performance,” Computers and Industrial Engineering, Vol.20, No.4, PP.461-468 (1991).
〔8〕Hadley, S. W., “Finding Part-machine Families Using Graph Partitioning Techniques,” International Journal of Production Research, Vol.34, No.7, PP.1821-1839(1996).
〔9〕Hark, H., and Sun, J. U., “A Genetic-Algorithm-Based Heuristic for the GT Cell Formation Problem,” Computers and Industrial Engineering, Vol.33, No.4, PP.941-955(1996).
〔10〕Joines, J. A., Manufacturing Cell Design Using Genetic Algorithms, Master Thesis dissertation, North Carolina State University(1996).
〔11〕King, J. R., “Machine-Component Grouping in Production Flow Analysis: An Approach Using Rank Order Clustering Algorithm,” International Journal of Production Research, Vol.18, No.2, PP.213-232(1980).
〔12〕Kirkpatrick, S., Gelatt, C. J., and Vecchi, M. P., “Optimization by Simulated Annealing,” Science, Vol.220, No.4598, PP.671-680(1983).
〔13〕Kusiak, A., and Cho, M., “The Generalized Group Technology Concept,” International Journal of Production Research, Vol.25, No.4, PP.561-569(1987).
〔14〕Marek, R., Witold P., “Evolutionary Optimization of Information Granules”, IEEE NAFIPS International Conference, PP.2035-2040(2001).
〔15〕McAuley, J., “Machine Grouping for Efficient Production,” Production Engineering, Vol.52, PP.53-57 (1972).
〔16〕McCormick, W. T. Jr., Schweitzer, P. J., and White, T. W., “Problem Decomposition and Data Reorganization by a Cluster Technique,” Operation Research, Vol. 20, No.5, PP.993-1009(1972).
〔17〕Michalewicz, Z., Genetic Algorithm + Date Structure = Evolution Programs, Springer Verlag, 2nd ed., Springer-Verlag, New York(1994).
〔18〕Moon, Y. B., and Chi, S. C., “Generalized Part Family Formation Using Neural Network Techniques,” Journal of Manufacturing System, Vo.l1, No.3, PP.149-159(1992).
〔19〕Purcheck, G. F. K., “A Mathematical Classification as a Basis, for the Design of Group Technology Production Cells,” Production Engineering, Vol.54, No.1, PP.35-46(1974).
〔20〕Rajagopalan, R., and Batra, J. L., “Design of Cellular Production Systems: A Graph Theoretic Approach,” International Journal of Production Research, Vol.13, No.6, PP.567-579(1975).
〔21〕Seifoddini, H., and M, D., “The Production Data-based Similarity
Coefficient versus Jaccard’s Similarity Coefficient,” Computers and Industrial Engineering, Vol.21, No.1-4, 263-266 (1991).
〔22〕Selim, H. M., Askin, R. G., and Vakharia, A. J., “Cell Formation in Group Technology: Review, Evaluation and Directions for Future Research,” Computers and Industrial Engineering, Vol.34, No.1, PP.3-20(1998).
〔23〕Srinivas M., “ Genetics Algorithms: A Survey ”, Computer, PP.17-26 , June(1994).
〔24〕Su, C. T., and Hsu, C. M., “A two-phase genetic algorithm for the cell formation problem”, International Journal of Industrial Engineering, Vol.3, No.2, 114-125 (1996).
〔25〕Suresh, N. C., Slomp, J., and Kaparthi, S., “The Capacitated Cell Formation Problem: A New Hierarchical Methodology,” International Journal of Production Research, Vol.33, No.6, PP.1761-1784(1995).
〔26〕Venugopal, V., and Narendran, T. T., “Cell Formation in Manufacturing Systems Through Simulated Annealing: An Experimental Evaluation,” European Journal of Operational Research, Vol.63, No.3, PP.409-422(1992).
〔27〕Wei, J. C., and Kern, G. M., “Commonality Analysis: a Linear Cell Clustering
Algorithm for Group Technology,” International Journal of Production Research, Vol.27, No.12, 2053-2062 (1989).
中文部份：
〔28〕王學文，「基因演算法在製造單元形成的應用」，國立台灣科技大學工業管理研究所碩士論文，民國86年。
〔29〕李忠贏，「以數學規劃法求解單元形成問題之最佳化模式及產生器之研究」，國立成功大學工業管理研究所碩士論文，民國87年。
〔30〕吳社邦主編，群組技術手冊CS-1分類與編號系統，中興管理顧問公司，民國74年。
〔31〕吳家瑭，「利用塔布搜尋法求解流程型工廠群組排程問題」，元智大學工業工程研究所碩士論文，民國87年。
〔32〕郭仲倫，「模糊自適應共振類神經網路於工件分族及機器分群上之應用：演算法之修正及效益評估」，元智工學院工業工程研究所碩士論文，國民84年。
〔33〕陳信嘉，「類神經網路應用於群組技術之機器分群及工件分族」，元智工學院工業工程研究所碩士論文，民國82年。
〔34〕鄧博文，「模糊集群神經網路模式之建立與應用」，義守大學管理科學研究所碩士論文，民國87年。