本研究的主要目的是以混成類神經網路為基本架構，建立醱酵系統之模式及其在進入工業化生產前的各項研究應用。希望能在最短時間及最少成本下，將零散的資料整合成大量生產所需的各種資訊。
混成類神經網路模式同時具有分析型及黑盒子型模式的優點。它的建立方法標準而簡單，對複雜的非線性系統具有良好描述能力，同時也可以被用來解釋系統行為，是醱酵系統理想的模式建立工具。此外，由於混成類神經網路模式的先驗（a priori）部份可隨系統實際情況修正，因此此種模式具有很強的韌性，可輕易加以修改。然而文獻中並沒有見到類似的應用，大部份的學者仍然將其視為一種改良型的類神經網路，僅僅使用於單一系統而未推廣至其他操作條件、環境或不同反應器之應用。本研究首先提出不同系統下的應用，利用操作較簡易、時間較短暫的批式實驗數據建立類神經網路，而先驗部份則可自由修改為饋料批式或連續式的系統方程式，使其成為饋料批式或連續式系統之模式。

傳統上進行醱酵系統穩定性分析所採用的相平面圖雖然簡易明瞭，但是需要進行多組連續式培養，成本高昂，失敗率也不低，本研究使用批式醱酵實驗建立的混成類神經網路模式來模擬連續式醱酵系統，並繪製相平面圖，使穩定性分析的成本大為減低。醱酵過程中常會因為混合不均勻導致局部酸鹼、溫度或溶氧偏差，利用混成類神經網路模式配合模式融合，可以準確地描述系統在環境擾動下的狀態變化。

如何有效地利用進料策略來提高饋料批式系統的產率亦是工業生產上常見的問題。傳統上的做法是進行離線最適化，然而這種做法容易因模式誤差或生產過程中的擾動導致效果不佳。因此，本研究提出半即時最適化策略，利用取樣數據重新進行最適化，以產生新的設定點來修正進料速度，效果十分良好。

至於非均勻混合反應器也是工業生產上常會遇到的系統，對於此種系統，只用單純的動力模式並無法有效地描述，應該將系統切割為生物部份及流力部份，分別建立其模式，最後再合併成一完整模式。本研究使用一組均勻混合反應器中的醱酵數據建立類神經網路，再配合反應曲線法所完成的槽序列模式，並加上氧氣質傳影響修正項，可以很完美地模擬網狀導流板氣舉式反應器中的細菌性纖維素（bacterial cellulose, BC）醱酵系統。

本論文利用混成類神經網路的架構，解決各項醱酵工業生產上所可能面對的問題。此種架構的好處在於模式建立程序標準而簡單、架構開放而具可修正性、模式準確度高，因此十分適合工業界之應用。

The main purposes of this study are to establish a hybrid modeling procedure of fermentation systems and to investigate its applications. It is desired to integrate scattered experimental data into useful information with less efforts and lower costs.
An HNN model possesses advantages of both analytical and black-box models. The modeling procedure is standardized and simple, and shows a good capacity to describe complex systems. In addition, an HNN model can be utilized to explain the behavior of a fermentation system. Therefore, it is an excellent modeling tool. Besides, the analytical part of an HNN model can be modified according to the actual conditions. The modeling method is flexible. However, no investigation has been reported to modify an existing HNN model to describe systems with different operating types, environmental conditions or reactors. In this study, we proposed the application of HNN models to simulate fermentation systems with different operating types. In the beginning, an ANN is established using the batch experimental data. After that, the system equations for a fed-batch or continuous system are utilized as the analytical part and combined with the ANN to give an HNN model. The HNN model can be employed to describe a fed-batch or continuous fermentation system according to the system equations.

Phase plane plots are frequently adopted to implement stability analyses of fermentation systems. However, quite a few continuous runs are required to make a phase plane plot and the costs are very high. In this study, batch runs are carried out to make an HNN model and the HNN model can be modified to simulate continuous runs. Several simulated runs can be obtained easily to make an accurate phase plane plot. By using the proposed method, the costs of the stability analyses are reduced substantially. On the other hand, the effects of imperfect mixing in a bioreactor on the environmental conditions are also considered. The imperfect mixing often leads to pH / temperature / DO deviation. It will be useful to predict the state variation due to the environmental change. Two HNN models in company with the model fusion method predict the state variation very well when a pH pulse occurs.

It is possible to improve the productivity of a fed-batch fermentation system by adjusting the feeding strategy. The conventional way is to carry out an off-line optimization and then follow the optimum feeding profile during the cultivation. Nevertheless, this method will be less effective when model errors or disturbances occur. Hence, we propose a semi-realtime optimization procedure to solve the problem. The semi-realtime optimization carries out a re-optimization procedure every one hour to determine new set-points. The new set-points are then employed to modify the feeding rate. The results are outstanding.

A fermentation system in an imperfect mixing reactor can be divided into the biological and hydrodynamic parts. An ANN is obtained from the experimental data of a cultivation in a stirred tank reactor. A tanks-in-series model is employed to describe the mixing properties of the fermentation system. The parameters are determined by a tracer response method. A modification according to the KLa is also employed. By combining these parts, a hybrid model is obtained to describe the fermentation system in a modified airlift reactor for bacterial cellulose production.

Due to the open structure and standardized modeling procedure of the HNN model, it is suitable for the modeling purposes on industrial production. The applications illustrated in this study form a template which is simple and flexible.

Adler, I., Deckwer, W. -D. and Sch*gerl, K., "Performance of Tower Loop Bioreactors - I, Model Calculations for Nonlimited Growth", Chemical Engineering Science, 37, 271-276, 1982a.
Adler, I., Deckwer, W. -D. and Sch*gerl, K., "Performance of Tower Loop Bioreactors - II, Model Calculations for Substrate Limited Growth", Chemical Engineering Science, 37, 271-276, 1982b.
Andre, G., Robinson, C. W. and Moo-Young, M., "New Criteria for Application of the Well-Mixed Model to Gas-Liquid Mass Transfer Studies", Chemical Engineering Science, 38, 1845-1854, 1983.
Andrews, J. F., "A Mathematical Model for the Continuous Culture of Microorganisms Utilizing Inhibitory Substrate", Biotechnology and Bioengineering, 10, 707-724, 1968.
Aynsley, M., Hofland, A., Morris, A. J., Montague, G. A. and DiMassimo, C., "Artificial Intelligence and the Supervision of Bioprocesses", Advanced Biochemical Engineering and Biotechnology, 48, 1-27, 1993.
Bagley, J. D., "The Behavior of Adaptive Systems which Employ Genetic and Correlation Algorithms", Doctoral Dissertation, University of Michigan, 1967.
Bailey, J. E., "Mathematical Modeling and Analysis in Biochemical Engineering : Past Accomplishments and Future Opportunities", Biotechnology Progress, 14, 8-20, 1998.
Bailey, J. E., "Mathematical Modeling and Analysis in Biochemical Engineering: Past Accomplishments and Future Opportunities", Biotechnology Progress, 14, 8-20, 1998.
Barricelli, N. A., "Numerical Testing of Evolution Theories", ACTA Biotheoretica, 16, 69-126, 1962.
Barricelli, N. A., "Symbiogenetic Evolution Processes Realized by Artificial Methods", Methodos, 9, 143-182, 1957.
Bischoff, K. B., "Accuracy of the Axial Dispersion Model for Chemical Reactors", AIChE Journal, 14, 820-821, 1968.
Blenke, H., "Loop Reactors", Advanced Biochemical Engineering, 13, 121-214, 1979.
Brauer, H., "Stirred Vessel Reactors", Biotechnology, 2, 395-444, Weinheim, New York, 1985.
Chisti, M. Y., "Airlift Bioreactors", Elsevier Applied Science, 1989.
Dhaouadi, H., Poncin, S., Hornut, J. M., Wild, G., Oinas, P. and Korpijarvi, J., "Mass Transfer in an External-Loop Airlift Reactor: Experiments and Modeling", Chemical Engineering Science, 52, 3909-3917, 1997.
Fields, P. R. and Slater, N. K. H., "Tracer Dispersion in a Laboratory Air-Lift Reactor", Chemical Engineering Science, 38, 647-653, 1983.
Foda, M. S., Salama, H. S. and Selim, M., "Factors Affecting Growth Physiology of Bacillus thuringiensis", Applied Microbiology and Biotechnology, 22, 50-52, 1985.
Fraser, A. S., "Simulation of Genetic Systems by Automatic Digital Computers. 5-Link-Age, Dominance and Epistasis. In O. Kempthorne (Ed.)", Biometrical Genetics, 70-83, New York, Macmillan, 1960.
Fraser, A. S., "Simulation of Genetic Systems", Journal of Theoretical Biology, 2, 329-346, 1962.
Fredrickson, A. G., McGee III, R. D. and Tsuchiya, H. M., "Mathematical Models in Fermentation Processes", Advanced Applied Microbiology, 23, 419, 1970.
Fu, P. -C. and Barford, J. P., "A Hybrid Neural Network - First Principles Approach for Modelling of Cell Metabolism", Computers and Chemical Engineering, 20, 6/7, 951-958, 1996.
Gavrilescu, M, Muntean, O. and Tudose, R. Z., "Stochastic Modeling of Axial Dispersion in External-Loop Airlift Bioreactors", Bioprocess Engineering, 23, 543-549, 2000.
Gavrilescu, M. and Tudose, R. Z., "Mixing Studies in External-Loop Airlift Reactors", Chemical Engineering Journal, 66, 97-104, 1997.
Goldberg, D. E., "Genetic Algorithms in Search, Optimization, and Machine Learning", Addison-Wesley Publishing Company, 1989.
Graefe, J., Bogaerts, P., Castillo, J. and Cherlet, M., "A New Training Method for Hybrid Models of Bioprocesses", Bioprocess Engineering, 21, 423-429, 1999.
Holland, J. H., "Adaptation in Natural and Artificial Systems", Ann Arbor: The University of Michigan Press, 1975.
Holland, J. H., "Outline for a Logical Theory of Adaptive Systems", Journal of the Association for Computing Machinery, 3, 297-314, 1962.
Hollstien, R. B., "Artificial Genetic Adaptation in Computer Control Systems", Doctoral Dissertation, University of Michigan, 1971.
Hwang, J. W., Yang, Y. K., Hwang, J. K., Pyun, Y. R. and Kim, Y. S., "Effects of pH and Dissolved Oxygen on Cellulose Production by Acetobacter xylinum BRC5 in Agitated Culture", Journal of Bioscience and Bioengineering, 88, 183-188, 1999.
Iguchi, M., Yamanaka, S. and Budhiono, A., "Bacterial Cellulose - a Masterpiece of Nature's Arts", Journal of Materials Science, 35, 261-270, 2000.
Kanai, T., Ichikawa, J. and Kawase, Y., "Dynamic Modeling and Simulation of Continuous Airlift Bioreactors", Bioprocess Engineering, 23, 213-220, 2000.
Kanai, T., Uzumaki, T. and Kawase, Y., "Simulation of Airlift Bioreactors: Steady-State Performance of Continuous Culture Processes". Computers and Chemical Engineering, 20, 1089-1099, 1996.
Kirkpatrick, S., Gelatt, C. D. and Vecchi, M. P., "Optimization by Simulated Annealing", Science, 220, 671-680, 1983.
Kohonen, T., "An Introduction to Neural Computing", Neural Networks, 1, 3-16, 1988.
Konstantinov, K. B., Yoshida, T. and Aarts, R., "Expert Systems in Bioprocess Control: Requisite Features", Advanced Biochemical Engineering and Biotechnology, 48, 169-191, 1993.
Kouda, T., Yano, H. and Yoshnaga, F., "Effect of Agitator Configuration on bacterial Cellulose Productivity in Aerated and Agitated Culture", Journal of Fermentation and Bioengineering, 83, 371-376, 1997.
Kr*lik, M., Ilavsk*, J., Pa*ek, J. and Lehock*, P., "Mathematical Dynamic Model of the Continuous Bubble Column Slurry Reactor", Chemical Engineering and Processing, 28, 127-132, 1990.
Luttmann, R., Thoma, M., Buchhloz, H. and Schugerl, K., "Model Development, Parameter Identification, and Simulation of the SCP Production Process in Air Lift Tower Bioreactors with External Loop - I. Generalized Distributed Parameter Model", Computers and Chemical Engineering, 7, 43-50, 1983.
Luttmann, R., Thoma, M., Buchhloz, H. and Schugerl, K., "Model Development, Parameter Identification, and Simulation of the SCP Production Process in Air Lift Tower Bioreactors with External Loop - II. Extended Culture Modeling and Quasi Steady Parameter Identification", Computers and Chemical Engineering, 7, 51-63, 1983a.
Martin, F. G. and Cockerham, C. C., "High Speed Selection Studies. In O. Kempthorne (Ed.)", Biometrical Genetics, 35-45, London, Pergamon Press, 1960.
Merchuk, J. C. and Stein, Y., "A Distributed Parameter Model for an Airlift Fermentor. Effects of Pressure", Biotechnology and Bioengineering, 23, 1309-1324, 1981.
Merchuk, J. C., Stein, Y. and Mateles, R. I., "Distributed Parameter Model of an Airlift Fermentor", Biotechnology and Bioengineering, 22, 1189-1211, 1980.
Michalewicz, Z., "Genetic Algorithms + Data Structures = Evolution Programs", Springer-Verlag, Berlin, 1992.
Mizobuchi, T., Morita, S. and Yano, T., "Stability and Phase-Plane Analyses of Continuous Phenol Biodegradation: A Simple Case", Journal of Fermentation Technology, 58, 33-38, 1980.
Monod, J., "The Growth of Bacterial Cultures", Annual Review of Microbiology, 3, 371-394, 1949.
Moriyama, H. and Shimizu, K., "On-Line Optimization of Culture Temperature for Ethanol Fermentation Using Genetic Algorithm", Journal of Chemical Technology and Biotechnology, 66, 217-222, 1996.
Narenda, K. S. and Parthasarathy, K., "Identification and Control of Dynamical Systems Using Neural Networks", IEEE Trans. Neural Networks, 1, 4-27, 1990.
O'Connor, G. M., Sanchez-Riera, F. and Cooney, C. L., "Design and Evaluation of Control Strategies for High Cell Density Fermentations", Biotechnology and Bioengineering, 39, 293-304, 1992.
Patnaik, P. R., "A Sufficiency Analysis of Asymptotic Stability in Continuous Recombinant Fermentation with Auxotrophic Mutants", Biochemical Engineering Journal, 2, 157-160, 1998.
Psichogios, D. C. and Ungar, L. H., "A Hybrid Neural Network-First Principles Approach to Process Modeling", AIChE Journal, 38, 10, 1499-1511, 1992.
Royce, P. N., "A Discussion of Recent Development in Fermentation Monitoring and Control from a Practical Perspective", Critical Reviews in Biotechnology, 13, 117-149, 1993.
Rubio, F. C., Garcia, J. L., Molina, E. and Chisti, Y., "Steady-State Axial Profiles of Dissolved Oxygen in Tall Bubble Column Bioreactors", Chemical Engineering Science, 54, 1711-1723, 1999.
Ruggeri, B. and Sassi, G., "On the Modelling Approaches of Biomass Behaviour in Bioreactor", Chemical Engineering Communications, 122, 1-56, 1993.
Schubert, J., Simutis, R., Dors, M., Havlik, I. and L*bbert, A., "Bioprocess Optimization and Control: Application of Hybrid Modelling", Journal of Biotechnology, 35, 51-68, 1994.
Shah, Y. T., Stiegel, G. J. and Sharma, M. M., "Backmixing in Gas-Liquid Reactors", AIChE Journal, 24, 369-400, 1978.
Shene, C., Diez, C. and Bravo, S., "Neural Networks for the Prediction of the State of Zymomonas Mobilis CP4 Batch Fermentations", Computers and Chemical Engineering, 23, 1097-1108, 1999.
Shi, Z. and Shimizu, K., "Neuro-Fuzzy Control of Bioreactor Systems with Pattern Recognition, Journal of Fermentation and Bioengineering, 74, 39-45, 1992.
Shih, C. C., Zuo, K. and Wu, W. W., "Optimization of Penicillin G Fed-Batch Fermentation via a Hybrid Neural Model and Real-Coded Genetic Algorithm", Proceedings of the 7th International Conference on Computer Applications in Biotechnology, Osaka, Japan, 51-54, 1998.
Shuler, M. L. and Kargi, F., "Bioprocess Engineering - Basic Concepts", Prentice-Hall International, 1992.
Simutis, R. and L*bbert, A., "A Comparative Study on Random Search Algorithms for Biotechnical Process Optimization", Journal of Biotechnology, 52, 245-256, 1997.
Simutis, R., Havlik, I. And L*bbert, A., "A Fuzzy-Supported Extended Kalman Filter: A New Approach to State Estimation and Prediction Exemplified by Alcohol Formation in Beer Brewing", Journal of Biotechnology, 24, 211-234, 1992.
Sugeno, M., "Industrial Applications of Fuzzy Control", North-Holland, Amsterdam, 1985.
Todt, J., Lucke, J.m Schugerl, K. and Renken, A., "Gas Holdup and Longitudinal Dispersion in Different Types of Multiphase Reactors and Their Possible Applications for Microbial Processes", Chemical Engineering Science, 23, 369-375, 1977.
Toyosaki, H., Naritomi, T., Seto, A., Matsuoka, M., Tsuchida, T. and Yochinaga, F., "Screening of Bacterial Cellulose-Producing Acetobacter Strain Suitable for Agitated Culture", Bioscience Biotechnology and Biochemistry, 59, 1498-1502, 1995.
Tung, H. L., Chiou, S. Y., Tu, C. C. and Wu, W. T., "An Airlift Reactor with Double Net Draft Tubes and Its Application in Fermentation", Bioprocess Engineering, 17, 1-5, 1997.
Turner, J. R. and Mills, P. L., "Comparison of Axial Dispersion and Mixing Cell Models for Design and Simulation of Fischer-Tropsch Slurry Bubble Column Reactors", Chemical Engineering Science, 45, 2317-2324, 1990.
Verlaan, P. and Tramper, J., "Hydrodynamics, Axial Dispersion and Gas-Liquid Oxygen Transfer in an Airlift-Loop Bioreactor with Three-Phase Flow", Bioreactors and Biotransformations, Elsevier, London, 363-373, 1987.
Verlaan, P., Tramper, J., van't Riet, K. and Luyben, K. Ch. A. M., "Hydrodynamics and Axial Dispersion in an Airlift-Loop Bioreactor with Two and Three-Phase Flow", Proceedings of the International Conference on Bioreactor Fluid Dynamics, Cambridge, U.K., 15-17, 1986a.
Verlaan, P., Tramper, J., van't Riet, K. and Luyben, K. Ch. A. M., "A Hydrodynamic Model for an Airlift-Loop Bioreactor with External Loop", Chemical Engineering Journal, 33, B43-B53, 1986b.
Wen, C. Y. and Fan, L. T., "Model for Flow Systems and Chemical Reactors", Marcel Dekker, New York, 1975.
Wilson, J. A. and Zorzetto, L. F., "A Generalised Approach to Process State Estimation Using Hybrid Artificial Neural Network/Mechanistic Models", Computers and Chemical Engineering, 21, 9, 951-963, 1997.
Wu, J. Y. and Wu, W. T., "Fed-Batch Culture of Saccharomyces cerevisiae in an Airlift Reactor with Net Draft Tube", Biotechnology Progress, 7, 230-233, 1991.
Wu, W. T. and Wu, J. Y., "Airlift Reactor with Net Draught Tube", Journal of Fermentation and Bioengineering, 70, 359-361, 1990.
Yang, Y. K., Park, S. H., Hwang, J. W., Pyun, Y. R. and Kim, Y. S., "Cellulose Production by Acetobacter Xylinum BRC5 under Agitated Condition ", Journal of Fermentation and Bioengineering, 85, 312-317, 1998.
Yang, Y. K., Park, S. H., Hwang, J. W., Pyun, Y. R. and Kim, Y. S., "Cellulose Production by Acetobacter xylinum BRC5 under Agitated Condition", Journal of Fermentation and Bioengineering, 85, 312-317, 1998.
Zuo, K. and Wu, W. T., "Semi-Realtime Optimization and Control of a Fed-Batch Fermentation System", Computers & Chemical Engineering, 24, 1105-1109, 2000.
Zuo, K. and Wu, W. T., "Prediction of Continuous Process Using Batch Data", Proceedings of the 4th Conference on Biochemical Engineering, Chiayi, Taiwan, 277-280, 1999.
Zuo, K., Wu, S. C., Cheng, H. P. and Wu, W. T., "Application of Model Fusion to State Prediction of a Fermentation System with pH value Variation" Proceedings of PSE Asia 2000, 369-373, Kyoto, Japan, 2000.
左克偉, "饋料批式醱酵之最適化 - 遺傳演算法與類神經網路之應用", 碩士論文, 國立清華大學, 1996.
江欽源, "以模擬退火配合類神經網路模式之實驗設計法", 碩士論文, 國立清華大學, 1995.
吳聲祺, "週期式操作連續式反應器：生化反應器之應用", 博士論文, 國立台灣大學, 1997.
易逸波, "類神經網路在離胺酸饋料批式醱酵製程中之線上狀態估計及控制之研究", 博士論文, 國立清華大學, 1995.
林正鄰, "模擬退火與類神經網路在製程最適化之應用", 博士論文, 國立清華大學, 1994.
花強and王樹青, "神經網絡建模方法在維生素C發酵過程中的應用", 化工學報, 47, 433-438, 1996.
施菁菁, "最適化饋料批式生產盤尼西林G之模擬分析", 碩士論文, 國立清華大學, 1997.
柯銀府, "蘇力菌醱酵系統之動態分析", 碩士論文, 國立清華大學, 2000.
熊定宇, "具網狀導流管氣舉式反應器之設計及其在蘇力菌醱酵上之應用", 博士論文, 國立清華大學, 1995.
鄭海鵬, "多重網狀導流板氣舉式反應器之規模放大及其在細菌纖維素生產上之應用", 碩士論文, 國立清華大學, 2001.
鍾建中, "網狀內管氣舉式反應器在蘇力菌素生產之應用", 博士論文, 國立清華大學, 1994.