512 Research Article Received: 22 September 2009 Accepted: 3 November 2009 Published online in Wiley Interscience: 3 February 2010 (www.interscience.wiley.com) DOI 10.1002/jctb.2319 Hybrid modeling of inulinase bio-production process Marcio A. Mazutti, a* Marcos L. Corazza, a Francisco Maugeri, b Maria I. Rodrigues, b J. Vladimir Oliveira, a Helen Treichel a and Fernanda C. Corazza c Abstract BACKGROUND: A potential application of inulinase in the food industry is the production of fructooligosaccharides (FOS) through transfructosilation of sucrose. Besides their ability to increase the shelf-life and flavor of many products, FOS have many interesting functional properties. The use of an industrial medium may represent a good, cost-effective alternative to produce inulinase, since the activity of the enzyme produced may be improved or at least remain the same compared with that obtained using a synthetic medium. Thus, inulinase production for use in FOS synthesis is of considerable scientific and technological appeal, as is the development of a reliable mathematical model of the process. This paper describes a hybrid neural network approach to model inulinase production in a batch bioreactor using agroindustrial residues as substrate. The hybrid modeling makes use of a series artificial neural network to estimate the kinetic parameters of the process and the mass balance as constitutive equations. RESULTS: The proposed model was shown to be capable of describing the complex behavior of inulinase production employing agroindustrial residues as substrate, so that the mathematical framework developed is a useful tool for simulation of this process. CONCLUSION: The hybrid neural network model developed was shown to be an interesting alternative to estimate model parameters since complete elucidation of the phenomena and mechanisms involved in the fermentation is not required owing to the black-box nature of the ANN used as parameter estimator. c  2010 Society of Chemical Industry Keywords: inulinase; hybrid modeling; stirred-tank; neural networks; complex systems; submerged INTRODUCTION A considerable amount of work has been carried out in recent years employing agroindustrial residues as substrate to produce inuli- nase by SSF 1–3 and by submerged fermentation. 4–6 This class of enzyme is of great importance in food processing, especially to syn- thesize fructooligosaccharides, which have numerous functional properties and are obtained from sucrose by transfructosilation, catalyzed by inulinase or β -fructosyltransferase. 7 Thus, the pro- duction of inulinase for use in FOS synthesis is of considerable importance, requiring the development of a reliable mathematical model of the process, an essential tool for scale-up, optimization and control and reactor design of inulinase production. An alternative for process modeling is the use of a hybrid formu- lation, aimed at including all available knowledge of the process. The foundations of hybrid models are the conservation principles: the poorly known or unknown properties of a process, such as reac- tion rate or model parameters, are estimated by an artificial neural network (ANN), including a priori knowledge of the process. 8–14 There are many different ways of making use of the hybrid modeling approach. Zorzeto et al. 15 presented two different configurations of hybrid model to outline the dynamics of a batch beer production. In the first, an ANN was used as an estimator of the reaction rate and in the second, the kinetic parameters were estimated. Teixeira et al. 16 described a model- based optimization of fed-batch BHK-21 cultures expressing the human fusion glycoprotein IgG1-IL2 using a hybrid model where the ANN was employed to correct Monod kinetic microbial growth. In previous work the authors developed a mathematical kinetic model for inulinase production using a phenomenological model and ANN model. 17 It was concluded that, due to the complexity of the medium composition and the complex dynamic behavior, it is rather difficult to use a phenomenological model with sufficient accuracy. To overcome this drawback, a methodology based on an ANN was adopted, with a 5-10-3 network. ∗ Correspondence to: Marcio A. Mazutti, Department of Food Engineering, URI - CampusdeErechim,Av.SetedeSetembro,1621-Erechim-RS,99700-000 – Brazil. E-mail: mazutti@uricer.edu.br a Department of Food Engineering, URI - Campus de Erechim, Av. Sete de Setembro, 1621-Erechim-RS, 99700-000 – Brazil b Department of Food Engineering, University of Campinas – UNICAMP, P.O Box 6121, CEP 13083-862, Campinas – SP, Brazil c Department of Chemical Engineering, Paran´ a Federal University (UFPR), Polytechnic Center, Jardim das Am´ ericas, Curitiba, PR, 82530-990, Brazil J Chem Technol Biotechnol 2010; 85: 512–519 www.soci.org c  2010 Society of Chemical Industry