本文提出一個以柔性演算法技術為基的最佳化策略，為利用預測信賴度導向於演化類神經模型搜尋(PREGSEN)，並結合有限實驗樣本的設計最佳化方法。利用實際系統的樣本訓練模擬類神經網路響應以進行最佳值搜尋，進而降低實驗成本。然而，由於實驗成本較高的關係，使用有限樣本是當前實際工程應用的現狀，但確可能阻礙了準確的模擬網路模型建立。少量的及偏頗的樣本分佈將導致複雜問題，缺乏模型的一般準確性。依據類神經網路模型的特性，預測準確度主要與預測設計和學習樣本的位置及距離存在著密切的關係。文中建議的最佳化策略是主張用於沿著設計最佳值迭代的演化型類神經網路模型，並建立模糊推論去估計網路模型的預測信賴度。引入預測信賴度以重新定義基因演算法的適應函數，導引於當前模擬網路去搜尋可信賴的暫時最佳值。將驗證過後的暫時最佳解當作一個額外的學習樣本，加入於先前的學習樣本中重新訓練類神經網路模型。直到最佳值收斂，網路的演化及搜尋迭代過程方才停止。網路模型將逐漸的改善，特別是在設計最佳值最可能存在的空間並提高其準確度，因而亦提升了樣本採樣的效率。提出兩個單極值最佳化範例與傳統迭代的類神經網路和遺傳演算法(NNGA)進行比較，來加以說明與驗證PREGSEN最佳化策略的效率與可行性。然後，再應用於汽車燃料儲存槽的製程最化設計及塑膠瓶的性能最佳化設計兩個擠出吹塑成型的工程問題，論證本文所提出PREGSEN策最佳化策略的優點。
多極值最佳化方面，本文提出結合下坡式簡單搜尋(DSS)法的多極值最佳化方法於模擬類神經網路搜尋暫時相對極值，並將每次驗證過後的暫時相對極值當作額外的學習樣本，加入於先前的學習樣本中重新訓練類神經網路模型。直到相對極值都到達收斂標準，網路的演化及搜尋迭代過程方才停止。以三個二變數多極值最佳化範例，驗證本文所提出方法的可行性及優越性。

This study proposes an optimization methodology, Prediction Reliability Guided Search of Evolving Network Modeling (PREGSEN), based on soft computing techniques, for design optimization with a limited number of experiments. Response samples from the actual system can be utilized to train a simulated neural network model prior carrying out an optimum search to reduce experimental costs. However, limited sample is status quo for engineering applications due to high experiment costs, which hamper the establishment of an accurate simulated model at once. Small and ill distributed samples will result in the lack of modeling generality for complex problems. In light of the characteristics of neural network model, the prediction accuracy is closely related to the location and the distance between the prediction design and the learning samples. The proposed scheme advocates an evolving network model along the iteration of design optimum. This study sets up fuzzy reasoning to estimate the prediction reliability of the network model. The prediction reliability is introduced to the definition of the fitness function of a genetic algorithm to guide the search for a reliable quasi-optimum of the current simulated model. The verification of the quasi-optimum serves as an additional sample to the previous learning samples to retrain the network model. The model evolution and searching processes iterate until the convergence of optimum. The network modeling improves gradually its precision especially in the most probable space of design optimum and thus enhances the sampling efficiency. Two benchmark numerical examples are presented to illustrate the feasibility and the efficiency compared with conventional iteration of NN and GA. Two engineering examples involving the extrusion blow molding of a gas tank and a bottle demonstrate the merits of the proposed scheme.
In the multimodal optimization, this study used the downhill simplex search method to search the relative quasi-optimum in the simulated neural network model. The verification of the relative quasi-optimum serves as the additional learning samples to the previous learning samples to retrain the neural network model. The model evolution and searching processes iterate until the convergence of the relative optimum. Two-variable of three benchmark numerical examples with multiple relative optimum are presented to illustrate the feasibility and the superiority of the proposed scheme.

[1]. G. Taguchi, “Performance Analysis Design”, International Journal of Production Research, v 16, n 6, 1978, p 521-530.
[2]. M. Kurt, E. Bagci, Y. Kaynak, “Application of Taguchi Methods in the Optimization of Cutting Parameters for Surface Finish and Hole Diameter Accuracy in Dry Drilling Progress” International Journal of Advanced Manufacturing Technology, v 40, n5-6, 2009, p458-469.
[3]. W. C. Chen, S. W. Hsu, “A Neural-Network Approach for an Automatic LED Inspection System”, Expert Systems with Applications, v 33, n 2, 2007, p 531-537.
[4]. H. L. Lin, T. Chou, C. P. Chou, “Optimization of Resistance Spot Welding Process Using Taguchi Method and a Neural Network”, Experimental Techniques, v 31, n 5, 2007, p 30-36.
[5]. N. H. Loh, S. C. Tam, S. Miyazzawa, “Use of Response Surface Methodology to Optimize the Finish in Ball Burnishing”, Precision Engineering, v 12, n 2, 1990, p 101-105.
[6]. G., Cybenko, “Approximation by Superpositions of a Sigmoidal Function”, Math. Control Signals Syst, v2, 1989, p303-314.
[7]. K. J. Cios, G. Y. Baaklini, A. Vary, “Soft Computing in Design and Manufacturing of Advanced Materials”, Journal of Engineering for Gas Turbines and Power, v 177, n 1, 1995, p 161-165.
[8]. M. Sanjari, K. Taheri, R. Movahedi, “An Optimization Method for Radial Forging Process Using ANN and Taguchi Method”, International Journal of Advanced Manufacturing Technology, v 40, n 7- 8, 2009, p 776- 784.
[9]. C. T. Su, C. C. Chiu and H. H. Chang, “Parameter Design Optimization via Neural Network And Genetic Algorithm”, Int. Journal Of Industrial Engineering: Theory Applications and Pratice, v7, n3, 2000, p224-231
[10]. A. K. Singh, S. S. Panda, D. Chakraborty, S. K. Pal, “Drill wear prediction using artificial neural network”, International Journal of Advanced Manufacturing Technology, v 28, n 5-6, 2006, p 456-462.
[11]. G. J. Wang, J. C. Tsai, P. C. Tseng, T. C. Chen, “Neural Taguchi Method for Robust Design Analysis”, Journal of the Chinese Society of Mechanical Engineers, v 19, n 2, 1998, p 223-230.
[12]. D. B. Fogel, “A Introduction to Simulated Evolutionary Optimization”, IEEE transductions on Neural Networks, v 5, n 1, 1994, p 3-14.
[13]. S. Santarelli, T. L. Yu, D. Goldberg, E. Altshuler, T. O’Donnell, H. Southall, R. Mailloux, “Military Antenna Design Using Simple and Competent Genetic Algorithms”, Mathematical and Computer Modelling, v 43, n 9-10, 2006, p 990-1022.
[14]. M. Pelikan, K. Sastry, D. Goldberg, “Scalability of the Bayesian Optimization Algorithm”, International Journal of Approximate Reasoning, v 31, n 3, 2002, p 221-258.
[15]. B. Nicolas, B. Pascal, “Optimization by hybridization of a genetic algorithm with constraint satisfaction techniques”, Proceedings of the IEEE Conference on Evolutionary Computation, 1998, p 645-649.
[16]. J. T. Tsai, T. K. Liu, J. H. Chou, “Hybrid Taguchi-Genetic Algorithm for Global Numerical Optimization”, IEEE Transactions On Evolutionary Computation, v 8, n 4, 2004, p365-377.
[17]. S. Nandi, S. Ghosh, S. S. Tambe, Kulkarni, B. D., “Artificial neural-network-assisted stochastic process optimization strategies”, AIChE Journal, v 47, n 1, 2001, p 126-141.
[18]. P. S. Chakravarthy, N. R. Babu, “A New Approach for Selection of Optimal Process Parameters in Abrasive Water Jet Cutting”, Materials and Manufacturing Processes, v 14, n 4, 1999, p 581-600.
[19]. H. Oktem, T. Erzurumlu, F. Erzincanli, “Prediction of Minimum Surface Roughness in End Milling Mold Parts Using Neural Network and Genetic Algorithm”, Materials and Design, v 27, n 9, 2006, p 735-744.
[20]. H. X. Huang and S. Lu, “Modeling Parison Formation in Extrusion Blow Molding by Neural Networks”, J. Appl. Polym. Sci., v 96, 2005, p 2230-2239.
[21]. T. Hanai, N. Iwata, T. Furuhashi, H. Honda, T. Kobayashi, “Proposal of Reliability Index in Search for Reliable Solution of Reverse Calculation Based on Fuzzy Neural Network Modeling”, Journal of Chemical Engineering of Japan, v 37, n 4, 2004, p 523-530.
[22]. J. Yu, X. Chen, T. R. Hung, F. Thibault, “Optimization of Extrusion Blow Molding Processes Using Soft Computing and Taguchi Method”, Journal of Intelligent Manufacturing, v 15, n 5, 2004, p 625-634.
[23]. Y. L. Hsu, Y. H. Dong, M. H. Hsu, “A sequential approximation method using neural networks for nonlinear discrete variable optimization with implicit constraints”, JSME International Journal, Series C, v 44, n 1, 2001, p 103-112.
[24]. H. X. Huang and S. Lu, “Neural Modeling of Parison Extrusion in Extrusion Blow Molding”, Journal of Reinforced Plastics and Composites, v 24, n 10, 2005, p 1025-1034.
[25]. R. W. Diraddo, A. Garcia-Rejon, “On-Line Prediction of Final Part Dimensions in Blow Molding: A Neural Network Computing Approach”, Polymer Engineering and Science, v 33, n 11, 1993, p 653-664.
[26]. D. Laroche, K. Kabanemi. L. Pecora, and R. Diraddo, “Integrated numerical modeling of the blow molding process”, Polymer Engineering and Science, v 39, n 7, 1999, p1223-1233.
[27]. J. Yu, T.R Hung, and F. Thibault, “Performance Optimization of Extrusion Blow Molded Parts Using Fuzzy Neural-Taguchi Method and Genetic Algorithm”, Proceedings of the ASME Design Engineering Technical Conference, v 2, 2002, p 133-139.
[28]. C. Gauvin, F. Thibault and D. Laroche, “Optimization of Blow Moulded Part Performance through Process Simulation”, Polymer Engineering and Science, v 43, n 7, 2003, p 1407-1414.
[29]. D. K. Lee and S. K. Soh, "Prediction of Optimal Preform Thickness Distribution in Blow Molding", Polymer Engineering and Science, v 36, n 11, 1996, p 1513-1520.
[30]. Y. L. Hsu, T. C. Liu, F. Thibault and B. Lanctot, “Design Optimization of the Blow Moulding Process Using a Fuzzy Optimization Algorithm”, Journal of Engineering Manufacture, Proc Instn Mech Engrs, Part B, v 218, n 2, 2004, p 197-212.
[31]. D. E. Goldberg and J. Richardson, “Genetic Algorithms with Sharing for Multimodal Function Optimization”, Proceedings of the 2nd International Conference on Genetic Algorithms, 1987, p 41-49.
[32]. K. Deb and D. E. Goldberg, “An Investigation of Niche and Species Formation in Genetic Function Optimization”, Proceedings of the 3rd International Conference on Genetic Algorithms, 1989, p 42-50.
[33]. 劉家有，“遺傳演算法為基礎之混合式全域最佳化方法”，臺灣工業技術學院機械工程學系碩士論文，1994。
[34]. 黃一鳴，“由已知設計點判斷近極值位置之多極值最佳化策略”， 臺灣工業技術學院機械工程學系碩士論文，1995。
[35]. 留英龍、顏沛華，“模糊理論應用於地層下陷之預測”， 臺灣工業技術學院機械工程學系碩士論文，1995。
[36]. 蘇木春、張孝德，機器學習：類神經網路、模糊系統以及基因演算法則，全華科技，1997。
[37]. 羅華強，類神經網路－MATLAB的應用，清蔚科技，2001。
[38]. 葉怡成，類神經網路模式應用與實作，儒林，2000。
[39]. BlowView, 8.0, © Copyright Industrial Materials Institute, National Research Council Canada, 2002.
[40]. L. A. Zadeh, “Fuzzy Sets”, Information and Control, v8, 1965, p 338-353.
[41]. 留英龍、顏沛華，“模糊理論應用於地層下陷之預測”，成功大學水利及海洋工程學系碩士論文，1999。
[42]. J. H. Holland, Adaptation in Natural and Artificial Systems, Ann Arbor, MI: The University of Michigan Press, 1975.
[43]. J. Nelder and R. Mead, “A Simplex Method for Function Minimization”, Computer Journal, v7, 1965, p 308-313.
[44]. 洪鵬翔，“中文新文自動群聚”， 清華大學資訊工程學系碩士論文，2000。
[45]. A. Karimi and P. Siarry, " Global Simplex Optimization - A simple and efficient metaheuristic for continuous optimization", Engineering Applications of Artificial Intelligence, v 25, n 1, 2012, p 48-55.
[46]. J. S. Jang, C. T. Sun, E. Mizutani, Neuro-Fuzzy and Soft Computing: a computational approach to learning and machine intelligence, Prentice-Hall, 1997.
[47]. J. Chen, D. S. H. Wong, S. S. Jang, S. L. Yang, “Product and Process Development Using Artificial Neural-Network Model and Information Analysis”, Journal of the AIChE, v 44, n 4, 1998, p 876-887.
[48]. D. Goldberg, Genetic Algorithms in Search, in: Optimization and Machine Learning, Addison-Wesley, Massachusetts, USA, 1989.
[49]. Y. W. Leung, Y. Wang, “An Orthogonal Genetic Algorithm with Quantization for Global Numerical Optimization”, IEEE Trans. On Evolutionary Computation, v 5, n 1, 2001, p 41-53.
[50]. S. Wimalin, T. James, “The optimisation of neural network parameters using Taguchi’s design of experiments approach: an application in manufacturing process modeling”, Neural Computing and Applications, v14, 2005, p 337-344.
[51]. D. Thierens, D. Goldberg, “Elitist Recombination: an integrated selection recombination GA”, International Conference on Evolutionary Computation, v 1, 1994, p 508-512.