TFT-LCD（Thin Film Transistor Liquid Crystal Display）產業分成薄膜電晶體陣列製作（TFT Array）、液晶面板組裝（LC Cell Assembly）與液晶面板模組組裝（Module Assembly）三大製程。目前台灣TFT-LCD產業蓬勃發展，在如此激烈的產業競爭環境下，除了在新技術上佔有先機外，如何在既有生產環境下改善週期時間（Cycle Time），即為產業所面臨的議題。
根據JIT的概念，小批量可以加速在製品的移動，加工時間隨著批量數減少而縮短，藉此達到縮短週期時間之目標，然而，TFT-LCD產業是藉由搬運設施進行搬運，其搬運產能會有所限制，換言之，移動批量就搬運績效與生產績效產生權衡關係。
本研究提出一適用於TFT-LCD的動態移動批量機制，此機制考慮所有系統環境因素，根據此環境的最佳移動批量大小，投入類神經網路進行訓練，以得到動態移動批量決策模組，根據系統環境之變化輸出決策值，定期調整移動批量，以求達到縮短週期時間之目標。
因此，本研究藉由離散事件模擬實際工廠運作情形，並導入動態移動批量機制後，比較固定移動批量，發現在各種情境下，本研究方法皆能有效改善週期時間，然而，降低移動批量卻不一定能夠有效改善，由此可知，本研究方法具備著穩健性。

TFT-LCD industry can be classified to three majority manufacturing procedure, including array, cell assembly and module assembly. TFT-LCD industry in Taiwan grows vigorously at present. Under such violent competition situation, besides take the advantages on the latest technology, we also need to know how to improve the cycle time in the inherent environment will be the anxious issue.
By the concept of JIT, small batch can smooth the moving of the WIP, and the process time will be reduced by the batch size to achieve the objection of curtailing cycle time. However, TFT-LCD industry always doing the moving functions by the automatic facility, it would have some restrictions for the moving capacity, in other words, there will have a trade-off between the moving performance and manufacturing performance of the moving size.
This paper proposed an appropriate moving sizing mechanics for the TFT-LCD industry. The mechanics considered all of the factors of the systematic environment. According the best moving batch size of this environment, would be inputted to train the neural network, and got a moving batch size module. According to the environment changing of the system and output the policy decision by neural network methods to modify the batch size, and achieving the objection of curtailing the process cycle time.
Consequently, our research simulated the realistic operation of the fabrication by discrete event simulation, and introduced the dynamic moving batch size mechanics. Compare to the fixed moving batch size, we find that our method can improve the performance in any situations. If we just reduced the moving batch size, we may not see the improvement, that proves our research have the characteristic of robust.

Photonic Industry and Technology Development Association，民97，全球光電市場與台灣產業之產值統計與分析，Available：http://www.pida.org.tw/temp/080109-newsrelease.doc
王淑珍，民 92，臺灣邁向液晶王國之秘中國生產力中心，台北縣。
葉怡成，民92，類神經網路模式應用與實作，第八版，儒林圖書，台北。
羅華強，民 94，類神經網路-MATLAB的應用，第七版，高立圖書，新竹。
顧鴻壽，民 93，光電液晶平面顯示器技術基礎及應用新文京開發，台北縣。
Adusumilli, K.M. and Wright, R.L., 2004, Comparative factory analysis of standard FOUP capacities, Proceedings of the 2004 Winter Simulation Conference, Austin, 2, 1930-1934.
Chen, M.C. and Yang, T., 2002, Design of manufacturing systems by a hybrid approach with neural network metamodelling and stochastic local search, International Journal of Production Research, 40, 71-92.
Chen, Q., Xi, L., and Wang, Y., 2006, The impact of release times, lot size, and scheduling policy in an A&T facility, The International Journal of Advanced Manufacturing Technology, 29, 577-583.
Chiang, W.C., Urban, T.L., and Baldridge, G. W., 1996, A neural network approach to mutual fund net asset value forecasting, Omega, 24, 205-215.
Davies, P.C., 1994, Design issues in neural network development, NEUROVEST Journal, 5, 21-25.
Gupta, V.K., Chen, J.G., and Murtaza, M.B., 1997, A learning vector quantization neural network model for the classification of industrial construction projects, Omega, 25, 715-727.
Hopp, W.J. and Spearman, M.L., 2001, Factory Physics, McGraw-Hill, New York.
Hush, D.R. and Horne, B.G., 1993, Progress in supervised neural networks waht's new since lippmann, IEEE Signal Processing Magazine, 10, 8-39.
Hwang, H., Kim, S.Y., and Moon, S.W., 1996, Determination of optimum unit load size of the AGV in an electronics assembly production system, International Journal of Production Research, 34, 1293-1306.
Jacobs, R.A., 1987, Increased Rates of Convergence Through Learning Rate Adaptation, Neural Networks, 1, 295-307.
Kelton, W.D., Sadowski, R.P., and Sturrock, D.T., 2007, Simulation with Arena, 4th edition, McGraw-Hill, New York.
Kim, B. and May, G.S., 1994, An optimal neural network process model for plasma etching, IEEE Transactions on Semiconductor Manufacturing, 7, 12-21.
Lessin, J., 1992, TFT acting up a storm onscreen, Computer Technology Review, 12, 6–30.
Lewis, C.D., 1982, Industrial and Business Forecasting Methods, Butterworth Scientific, London.
Liao, H.Y., Liu, S.J., Chen, L.H., and Tyan, H.R., 1995, A bar-code recognition system using backpropagation neural networks, Engineering Applications of Artificial Intelligence, 8, 81-90.
Martinez, S.E., Smith, A.E., and Bidanda, B., 1994, Reducing waste in casting with a predictive neural model, Journal of Intelligent Manufacturing, 5, 277-286.
McCulloch, W.S. and Pitts, W., 1943, A logical calculus of the ideas immanent in nervous activity, Bulletin of Mathematical Biology, 5, 115-133.
Medsker, L. and Liebowtiz, J., 1994, Design and Development of Expert Systems and Neural Networks, Macmillan, New York.
Moon, S.W. and Hwang, H., 1999, Determination of unit load sizes of AGV in multi-product multi-line assembly production systems, International Journal of Production Research, 37, 3565-3581.
Nicholas, J. and Soni, A., 2005, The Portal to Lean Production: Principles and Practices for Doing More with Less, Auerbach Publications, New York.
Schmeiser, B., 1982, Batch size effects in the analysis of simulation output, Operations Research, 30, 556-568.
Stevenson, W.J., 2002, Operations management, 7th edition, McGraw-Hill/Irwin, Boston.
Su, C.T., Yang, T., and Ke, C.M., 2002, A neural-network approach for semiconductor wafer post-sawinginspection, IEEE Transactions on Semiconductor Manufacturing, 15, 260-266.
Tam, K.Y. and Kiang, M.Y., 1992, Managerial applications of neural networks: the case of bank failure predictions, Management Science, 38, 926-947.
Walczak, S., 2001, An empirical analysis of data requirements for financial forecasting with neural networks, Journal of Management Information Systems, 17, 203-222.
Wang, C.N. and Wang, C.H., 2007, A simulated model for cycle time reduction by acquiring optimal lot size in semiconductor manufacturing, The International Journal of Advanced Manufacturing Technology, 34, 1008-1015.
Watrous, R.L., 1987, Learning algorithms for connectionist networks: applied gradient methods of nonlinear optimization, Proceedings of the IEEE International Conference on Neural Networks, 2, 619-627.
Yang, T., Lin, H.C., and Chen, M.L., 2006, Metamodeling approach in solving the machine parameters optimization problem using neural network and genetic algorithms: a case study, Robotics and Computer Integrated Manufacturing, 22, 322-331.
Yang, T., Tsai, T.N., and Yeh, J., 2005, A neural network-based prediction model for fine pitch stencil-printing quality in surface mount assembly, Engineering Applications of Artificial Intelligence, 18, 335-341.
Zhang, G., Eddy Patuwo, B., and Y. Hu, M., 1998, Forecasting with artificial neural networks: the state of the art, International Journal of Forecasting, 14, 35-62.
Zuperl, U. and Cus, F., 2003, Optimization of cutting conditions during cutting by using neural networks, Robotics and Computer Integrated Manufacturing, 19, 189-199.