Downloads 290

..............................

Views 1k

..............................

Cited by

..............................

Received date November, 28 2007

Accepted date April 24, 2008 1. I

Fuzzy and Neuro-Fuzzy Modeling of a Fermentation Process

Author Abstract: Neuro-fuzzy modeling may be qualified as a grey-box technique, since it combines the transparency of rule-based ,

Keywords #Yeast fermentation # fed-batch # takagi-sugeno model # levenberg-marquardt algorithm

Abstract Neuro-fuzzy modeling may be qualified as a grey-box technique, since it combines the transparency of rule-based fuzzy systems with the learning capability of neura l networks. The main problem in the identification of non-linear processes is the lack of complete information. Certain variables are, either immeasurable or difficult to measure, the soft sensors are the necessary tools to solve the problem. Those latter can be used via online estimation, and then they wi ll be implemented in fed- batch fermentation processes for optimal production and online monitoring. The process parameters are estimated through a fuzzy logic system. The fuzzy models of takagi-suge no type suffer of the problem of poor initialization, which can be solved by the trial-and error method Trial-and-error method i s used to solve the poor initialization problem of TS models, this deals with identifying the structure of the model, such struct ure consists on finding the optimum number of rules , which enters in the model cost reduction. The fuzzy model might not cap ture the process non-linearity, especially if the number of rules is over- optimized. Bioreactors exhibit a wide range of dyna mic behaviours and offer many challenges to modelin g, as a result of the presence of living micro-organisms whose growth rat e is described by complex equations. We will illustrate the fuzzy and the neuro-fuzzy modeling on the identification of such a system. In order to compare the NF model outputs, we use another fuzzy model that does not incorporate the neural network learning capability, to identify the parameters of the same process. Even though, the two models were trained using levenberg-marquardt algorithm, the corresponding simulation results show that a better modeling is achieved using NF technique, esp ecially that we did not employ any involved optimization procedure to identify the NF structure.

References

[1] Bellgardt H., Application of an EKF for State Estimation of a Yeast Fermentation, in IEE Proceedings , Germany, pp. 226*234, 1986.

[2] Celina L. and Filomena S., Heuristic Sensitivity Analysis of Baker s Yeast Model Parameters, in Proceedings of Investigacio Operacional, Portugal, pp. 247*263, 2004. 384 The International Arab Journal of Information Technology, Vol. 6, No. 4, October 2009

[3] Euntai K., A Transformed Input*Domain Approach to Fuzzy Modeling, Computer Journal of IEEE Transactions on Fuzzy System , vol. 6, no. 4, pp. 596*604, 1998.

[4] Jang R., ANFIS: Adaptive Network Based Fuzzy Inference System, in Proceedings of IEEE Transactions on Syststem Man Cybeern, USA, pp. 665*685, 1993.

[5] Oliveira R., Supervision, Control and Optimization of Biotechnological Processes, PhD Thesis, Martin*Luther University, Portugal, 1998.

[6] Panchariya C., Nonlinear System Identification Using Takagi*Sugeno Type Neuro*Fuzzy Model, in Proceedings of the Second IEEE International Conference on Intelligent Systems , India, pp. 76*81, 2004.

[7] Takagi T. and Sugeno M., Fuzzy Identification of Systems and its Applications to Modeling and Control, in Proceedings of IEEE Transactions on Syststem Man Cybern, Japan, pp. 116*132, 1985.

[8] Tzafestas G. and Zikidis C., On*Line Neuro Fuzzy ART Based Structure and Parameter Learning TSK Model, in Proceedings of IEEE Transactions on Syststem Man Cybeern , Greece, pp. 797*802, 2001.

[9] Xiong Z., On Line Estimation of Concentration Parameters in Fermentation Processes, Journal of Zhejiang University SCIENCE, vol. 6, no. 3, pp. 530*534, 2005. Chabbi Charef received his BSc in electrical engineering from USTO University, Algeria in 1981, his MSc from Ohio University, USA in 1985, and currently, preparing his PhD in electrical engineering. Currently, he is an assistant professor at BadjiMokhtar University, Algeria since 1988. Mahmoud Taibi received his BSc in electrical engineering from USTO University, Algeria in 1980, and his MSc from Badji*Mokhtar University, Algeria in 1996. Currently, he is an assistant professor at Badji*Mokhtar University, Algeria since 1983. His interests are i n intelligent systems. Nicole Vincent is full professor since 1996. She presently heads the Center of Research in Computer Science (CRIP5) and the team Syst mes Intelligents de Perception (SIP) in the University Paris Descartes, France. After studying in Ecole Normale Sup rieure and graduation in mathematics, and graduation in mathematics, Nicole Vincent received a PhD in computer science in 1988 from Lyon Insa. 385 The International Arab Journal of Information Technology, Vol. 6, No. 4, October 2009 386 The International Arab Journal of Information Technology, Vol. 6, No. 4, October 2009 Fuzzy and neuro-fuzzy modeling of a fermentation Process

Abstract: Neuro-fuzzy modeling may be qualified as a grey-box technique, since it combines the transparency of rule-based fuzzy systems with the learning capability of neura l networks. The main problem in the identification of non-linear processes is the lack of complete information. Certain variables are, either immeasurable or difficult to measure, the soft sensors are the necessary tools to solve the problem. Those latter can be used via online estimation, and then they wi ll be implemented in fed- batch fermentation processes for optimal production and online monitoring. The process parameters are estimated through a fuzzy logic system. The fuzzy models of takagi-suge no type suffer of the problem of poor initialization, which can be solved by the trial-and error method Trial-and-error method i s used to solve the poor initialization problem of TS models, this deals with identifying the structure of the model, such struct ure consists on finding the optimum number of rules , which enters in the model cost reduction. The fuzzy model might not cap ture the process non-linearity, especially if the number of rules is over- optimized. Bioreactors exhibit a wide range of dyna mic behaviours and offer many challenges to modelin g, as a result of the presence of living micro-organisms whose growth rat e is described by complex equations. We will illustrate the fuzzy and the neuro-fuzzy modeling on the identification of such a system. In order to compare the NF model outputs, we use another fuzzy model that does not incorporate the neural network learning capability, to identify the parameters of the same process. Even though, the two models were trained using levenberg-marquardt algorithm, the corresponding simulation results show that a better modeling is achieved using NF technique, esp ecially that we did not employ any involved optimization procedure to identify the NF structure.
URL: https://iajit.org/paper/4435

,abstract={ Neuro-fuzzy modeling may be qualified as a grey-box technique, since it combines the transparency of rule-based fuzzy systems with the learning capability of neura l networks. The main problem in the identification of non-linear processes is the lack of complete information. Certain variables are, either immeasurable or difficult to measure, the soft sensors are the necessary tools to solve the problem. Those latter can be used via online estimation, and then they wi ll be implemented in fed- batch fermentation processes for optimal production and online monitoring. The process parameters are estimated through a fuzzy logic system. The fuzzy models of takagi-suge no type suffer of the problem of poor initialization, which can be solved by the trial-and error method Trial-and-error method i s used to solve the poor initialization problem of TS models, this deals with identifying the structure of the model, such struct ure consists on finding the optimum number of rules , which enters in the model cost reduction. The fuzzy model might not cap ture the process non-linearity, especially if the number of rules is over- optimized. Bioreactors exhibit a wide range of dyna mic behaviours and offer many challenges to modelin g, as a result of the presence of living micro-organisms whose growth rat e is described by complex equations. We will illustrate the fuzzy and the neuro-fuzzy modeling on the identification of such a system. In order to compare the NF model outputs, we use another fuzzy model that does not incorporate the neural network learning capability, to identify the parameters of the same process. Even though, the two models were trained using levenberg-marquardt algorithm, the corresponding simulation results show that a better modeling is achieved using NF technique, esp ecially that we did not employ any involved optimization procedure to identify the NF structure. },
keywords={Yeast fermentation, fed-batch, takagi-sugeno model, levenberg-marquardt algorithm},
ISSN={2413-9351},
month={Jan}}

AB - Neuro-fuzzy modeling may be qualified as a grey-box technique, since it combines the transparency of rule-based fuzzy systems with the learning capability of neura l networks. The main problem in the identification of non-linear processes is the lack of complete information. Certain variables are, either immeasurable or difficult to measure, the soft sensors are the necessary tools to solve the problem. Those latter can be used via online estimation, and then they wi ll be implemented in fed- batch fermentation processes for optimal production and online monitoring. The process parameters are estimated through a fuzzy logic system. The fuzzy models of takagi-suge no type suffer of the problem of poor initialization, which can be solved by the trial-and error method Trial-and-error method i s used to solve the poor initialization problem of TS models, this deals with identifying the structure of the model, such struct ure consists on finding the optimum number of rules , which enters in the model cost reduction. The fuzzy model might not cap ture the process non-linearity, especially if the number of rules is over- optimized. Bioreactors exhibit a wide range of dyna mic behaviours and offer many challenges to modelin g, as a result of the presence of living micro-organisms whose growth rat e is described by complex equations. We will illustrate the fuzzy and the neuro-fuzzy modeling on the identification of such a system. In order to compare the NF model outputs, we use another fuzzy model that does not incorporate the neural network learning capability, to identify the parameters of the same process. Even though, the two models were trained using levenberg-marquardt algorithm, the corresponding simulation results show that a better modeling is achieved using NF technique, esp ecially that we did not employ any involved optimization procedure to identify the NF structure.