Application of model-predictive control based on artificial neural networks to optimize the fed-batch process for riboflavin production.

The fed-batch process for commercial production of riboflavin (vitamin B2) was optimized on-line using model-predictive control based on artificial neural networks (ANNs). The information required for process models was extracted from both historical data and heuristic rules. After each cultivation the process model was readapted off-line to include the most recent process data. The control signal (feed rate), however, was optimized on-line at each sampling interval. An optimizer simulated variations in the control signal and assessed the forecasted model outputs according to an objective function. The optimum feed profile for increasing the product yield (YB2/S) and the amount of riboflavin at the time of harvesting was adjusted continuously and applied to the process. In contrast to the control by set-point profiles, the novel ANN-control is able to react on-line to variations in the process and also to incorporate the new process information continuously. As a result, both the total amount of riboflavin produced and the product yield increased systematically by more than 10% and the reproducibility of seven subsequently optimized batches was enhanced.

[1]  W. Fred Ramirez,et al.  Optimization of Fed‐Batch Bioreactors Using Neural Network Parameter Function Models , 1996 .

[2]  David W. Clarke,et al.  Generalized predictive control - Part I. The basic algorithm , 1987, Autom..

[3]  J Horiuchi,et al.  Industrial application of fuzzy control to large-scale recombinant vitamin B2 production. , 1999, Journal of bioscience and bioengineering.

[4]  H J van Can,et al.  An efficient model development strategy for bioprocesses based on neural networks in macroscopic balances. , 1997, Biotechnology and bioengineering.

[5]  N. Hannett,et al.  Genetic engineering of Bacillus subtilis for the commercial production of riboflavin , 1999, Journal of Industrial Microbiology and Biotechnology.

[6]  Henry C. Lim,et al.  Control of Bioreactor Systems , 2001 .

[7]  Robert Hooke,et al.  `` Direct Search'' Solution of Numerical and Statistical Problems , 1961, JACM.

[8]  J Morris,et al.  Neural-network contributions in biotechnology. , 1994, Trends in biotechnology.

[9]  Stefan Gehlen.  Untersuchungen zur wissensbasierten und lernenden Prozessführung in der Biotechnologie , 1993 .

[10]  A. Johnson,et al.  The control of fed-batch fermentation processes - A survey , 1987, Autom..

[11]  Peter A. Vanrolleghem,et al.  Rolling learning-prediction of product formation in bioprocesses , 1999 .

[12]  W. Bretzel,et al.  Commercial riboflavin production by recombinant Bacillus subtilis: down-stream processing and comparison of the composition of riboflavin produced by fermentation or chemical synthesis , 1999, Journal of Industrial Microbiology and Biotechnology.

[13]  J. A. Roels,et al.  Energetics and Kinetics in Biotechnology , 1983 .

[14]  U. Sauer,et al.  Physiology and metabolic fluxes of wild-type and riboflavin-producing Bacillus subtilis , 1996, Applied and environmental microbiology.

[15]  Kevin Warwick,et al.  Dynamic Systems in Neural Networks , 1995 .

[16]  David W. Clarke,et al.  Generalized Predictive Control - Part II Extensions and interpretations , 1987, Autom..

[17]  Jayant M. Modak,et al.  Optimization of fed-batch bioreactor using neural network model , 1998 .

[18]  Douglas B. Kell,et al.  GMP — good modelling practice: an essential component of good manufacturing practice , 1995 .

[19]  James S. Albus,et al.  New Approach to Manipulator Control: The Cerebellar Model Articulation Controller (CMAC)1 , 1975 .

[20]  F.-S. Wang,et al.  Optimal feed policy for fed-batch fermentation of ethanol production by Zymomous mobilis , 1997 .

[21]  K. Kovar,et al.  ARTIFICIAL NEURAL NETWORKS FOR ON-LINE OPTIMISATION OF BIOTECHNOLOGICAL PROCESSES , 1999 .

[22]  Tsuneo Yamane,et al.  Fed-batch techniques in microbial processes , 1984 .

[23]  M. Ott,et al.  Molecular cloning and characterisation of the ribC gene from Bacillus subtilis : a point mutation in ribC results in riboflavin overproduction , 1997, Molecular and General Genetics MGG.

[24]  Rimvydas Simutis,et al.  Bioprocess optimization and control: Application of hybrid modelling , 1994 .

[25]  G. F. Andrews.  Bioprocess optimization and control: Application of hybrid modelling (Schubert et al. J. Biotechnol. 35 (1994) 51–68) , 1995 .

[26]  Manel Poch,et al.  Biomass Estimation in Plant Cell Cultures: A Neural Network Approach , 1995 .

[27]  Kenneth J. Hunt,et al.  Neural Adaptive Control Technology , 1996 .

[28]  M. Thoma,et al.  Neurocontrol: Learning Control Systems Inspired by Neuronal Architectures and Human Problem Solving Strategies , 1992 .

[29]  T. Egli,et al.  Growth Kinetics of Suspended Microbial Cells: From Single-Substrate-Controlled Growth to Mixed-Substrate Kinetics , 1998, Microbiology and Molecular Biology Reviews.

[30]  P. Teissier,et al.  A hybrid recurrent neural network model for yeast production monitoring and control in a wine base medium , 1997 .

[31]  K. Yamuna Rani,et al.  Control of fermenters : a review , 1999 .

[32]  A. Bacher,et al.  GTP cyclohydrolase II and 3,4-dihydroxy-2-butanone 4-phosphate synthase are rate-limiting enzymes in riboflavin synthesis of an industrial Bacillus subtilis strain used for riboflavin production , 1999, Journal of Industrial Microbiology and Biotechnology.

[33]  H. Hohmann,et al.  Development of a Fermentation Process for the Manufacture of Riboflavin , 1996, CHIMIA.

[34]  Rimvydas Simutis,et al.  Using measurement data in bioprocess modelling and control , 1994 .

[35]  Walter Sebastian Mischo.  How to Adapt in Neurocontrol: A Decision for CMAC , 1996 .