近年來由於生產型態的變遷，加上生產技術的提升，使得生產管理的工作日益複雜。尤其是面對工廠中大量的訂單、龐大的機台數、繁複的加工程序，造成許多不確定性的情形。在傳統的存貨政策中，通常企業考量的是單階庫存(Installation Inventory)，但經常無法真正的反應整個製程內部存量的動態行為。自從供應鏈管理成為重要研究領域後，多階庫存(Echelon Inventory)便成為重要的存貨政策，且現今生產系統都已資訊化，也擁有相當豐富之生產資訊。
　　
　　柔性計算之一般性定義包括模糊集合、類神經網路與遺傳演算法等的計算方法。常使用於解決資料探勘的問題，除了致力於提供低成本的近似解之外，也加速了運算的過程，對大量資料處理與複雜問題的求解有相當不錯的表現。模糊集合常應用於分群、關聯法則、時間序列分析與影像檢索；類神經網路乃由規則中萃取知識、量化評估、分群與分類；而遺傳演算法則提供有效率的最佳化搜尋機構，常應用於迴歸與關聯法則。
　　
　　故本研究將生產系統之狀態變數以基因演算法進行特徵選取，使用模糊類神經網路將生產系統輸出入狀態變數之關係以模糊推論邏輯表達，分別將資料集進行訓練與測試，評估分類準確率並與其他分類工具進行比較。再以此模糊控制器作為生產系統存量管制控制器，分析生產系統在此種控制方式下之生產成本變化，最後，針對經濟訂購量進行敏感度分析。
　　
　　本研究之研究結論分為兩部分。第一部份以生產決策作為輸出變數，得知適應性模糊推論系統除了接近類神經網路之高分類準確率外，亦比同樣能產生決策法則之決策樹準確率來的高，讓初學者即使在成本資訊不充分的狀況下，也能根據模糊規則庫進行精準的現場決策判斷。第二部份以經濟批量係數作為輸出變數，得知適應性模糊推論系統之訓練結果可放寬整數倍批量係數的前提假設。

　　Recently, in response to the changes of production type and improvement of production skills, production management becomes more and more complicated. Besides, a variety of orders, machines, and process also causes a lot of uncertainty. Conventional inventory policy considers only installation inventory policy. However, it doesn’t usually reflect the dynamic behavior inside business process completely. After supply chain management become main research field, echelon inventory has played an important role in inventory policy.
　　
　　Soft computing is generally defined as all approaches including fuzzy sets, neural network (NN), and genetic algorithm (GA). It usually used to solve data mining problems and provide approximation for lowering cost. The application tool not only accelerates calculating process, but also performs excellently at dealing with complicated situation which contains huge datasets. Fuzzy set is often applied to clustering, association rule, time series analysis, and image retrieval. Neural network extracts knowledge from rules, and used to evaluation, clustering, and classification. In addition to these methods, genetic algorithm is an efficient search engine for optimization and frequently applied in regression and association rule.
　　In this study, genetic algorithm is used to perform feature selection among production variables. Then, apply fuzzy-neural network to express fuzzy inference logic for input-output relations and perform training and testing on three datasets respectively. In the next step, classification accuracy is evaluated and compared with other classification tools. Finally, think of the fuzzy controller as production inventory controller and analysis the changes of production cost. Furthermore, perform sensitivity analysis for economic order quantity (EOQ).
　　
　　The research reached two major conclusions. First, production decision was used as the output variable. It was found that Adaptive-Network-Based Fuzzy Inference System (ANFIS) not only reached high classification accuracy as NN, but also performed better than decision tree. Beginners could make correct decision even under insufficient information. In the rest part, the coefficient of economic batch quantity was used as the output variable. The training results of ANFIS relax the cost equations’ premise suggested by De Bodt, Graves and Mitra, Chatterjee, on assuming that the order quantity at stage 2 is an integer multiple of Q1. i.e., Q2=N*Q1, where N is an integer.

一、 中文部份
葉怡成(1996)，類神經網路模式應用與實作，儒林書局。

張斐章、張麗秋、黃浩倫(2004)，類神經網路理論與實務，台北市：東華書局。

王瑞瑜(2005)，資料探勘方法在兩級生產系統之存量管制應用，成功大學工業與資訊管理學系專班，未出版碩士論文。

二、 英文部分
Axsäter S. (1993) “Continuous Review Policies for Multi-Level Inventory Systems with Stochastic Demand.” In: Graves SC, et al., editor. Handbook in OR & MS, vol. 4. Amsterdam: Elsevier Science Publishers, pp.175-97.

Axsater, S. and K. Rosling (1993) “Note: Installation vs. Echelon Stock Policies for Multilevel Inventory Control.” Management Science, 39(10), pp.1274-1280.

Boritz, J. and Kennedy, D.(1995), “Effectiveness of Neural Network Types for Prediction of Business Failure.” Expert Systems with Application, 9(4), pp.503-512.

Chen, F. and Zheng Y. (1994) “Lower Bounds for Multi-Echelon Stochastic Inventory System.” Management Science, 40(11), pp.1426-1443.

Cristianini, N. and Shawe-Taylor, J.(2000), “An Introduction to Support Vector Machines.” Cambridge University Press, Cambridge, UK.

Clark, A. J., and Scarf, H. (1960) “Optimal Policies for a Multi-Echelon Inventory Problem.” Management Science, pp.175-190.

Dash, M. and Liu, H.(1997), “Feature Selection for Classification.” Intelligent Data Analysis, pp.131-156.

De Bodt, M. A. and Grave S. C. (1985) “Continuous Review Policies for a Multi-Echelon Inventory Problem with Stochastic Demand.” Management Science, 31(10), pp.1286-1299.

Duhja, M. and Morko, G. (2003), “Feature Selection on Hierarchy of Web Documents.” Decision Support Systems, 35(1), pp.45-87.

Federgruen, A. and Zipkin, P. (1984) “Computational Issues in An Infinite-Horizon, Multi-Echelon Inventory Model.” European Journal of Operational Research, pp.818-836.

Federgruen A. (1993) “Centralized Planning Models for Multi-Echelon Inventory Systems under Uncertainty.” In: Graves SC, et al., editor. Handbook in OR & MS, vol. 4. Amsterdam: Elsevier Science Publishers, pp.133-73.

Freeman, J. A. and Skapura D. M. (1992) “Neural Networks Algorithms, Applications and Programming Techniques.” Addison-Wesley, New York.

Gately, E.(1996), Neural Networks for Financial Forecasting. John Wiley & Sons, Canada.

Hall, M. A. (1998) “Correlation-based Feature Selection for Machine Learning.” Ph.D Dissertation. Hamilton, NZ: Waikato University, Department of Computer Science.

Holland, J. H. (1975) “Adaptation in Natural and Artificial Systems.” University of Michigan Press, Ann Arbor, MI.

John, G. H., Kohavi, R. and Pfleger, K.(1994), “Irrelevant Features and the Subset Selection Problem, ” Proceedings of the Eleventh International Conference on Machine Learning, pp.121-129.

Jang J. S. R (1993) “ANFIS: Adaptive Network-Based Fuzzy Inference System.” IEEE Transactions on Neural Networks. 3(5), pp.714-723.

Lewis, D. D.(1992), “Feature Selection and Feature Extraction For Text Categorization,” Proceedings of Speech and Natural Language Workshop, pp.212-217.

Matta, K. F. and Sinha D. (1995) “Policy and Cost Approximations of Two-Echelon Distribution Systems with a Procurement Cost at the Higher Echelon.” IEEE Transactions, 27, pp.638-645.

Mitra, S., Chatterjee, A. K. (2004) “Echelon Stock Based Continuous Review (R, Q) Policy for Fast Moving Items.” Management Science, pp.161-166.

Mitra, S., Pal, S. K., and Mitra, P. (2002) “Data mining in soft computing framework: A survey.” IEEE Transactions on Neural Networks. vol. 13, No. 1, pp.3-14.

Silver, E. A. and Pyke, D. F., and Peterson, R. (1998) Inventory Management and Production Planning and Scheduling. Wiley.

Sinha, D. and Matta K. F. (1991) “Multi-echelon (R, S) Inventory Model.” Decision Sciences, 22(3), pp484-499.

Svoronos, A. and P. Zipkin (1988) “Estimating the Performance of Multilevel Inventory Systems.” European Journal of Operational Research, 36(1), pp57-72.

Tickle, A.B., Andrews, R., Golea, M., and Diederich, J. (1998) “The Truth Will Come to Light: Directions and Challenges in Extracting the Knowledge Embedded within Trained Artificial Neural Networks.” IEEE Transactions on Neural Networks. Vol.9, pp1057-1068.

Zadeh, L. A. (1994) “Fuzzy Logic, Neural Networks, and Soft Computing.” Communications of ACM, vol. 37, pp.77-84.

Zadeh, L. A. (1965) “Fuzzy sets”, Information and Control, vol. 8, pp.338-353.