Nonlinear Identification of Recombinant E. coli Fed-Batch Fermentation with a Hybrid Method

Mahdi Feyzdar, Valiollah Babaeipure and Ahmad Reza Vali



The paper deals with the identification of a fed-batch biotechnological process based on experimental data from process. The process considered in the study is fed-batch cultivation of E. coli BL21 (DE3) [pET3a-ifn_] under maximum attainable specific growth rate for producing _-Interferon protein that maximum cell density (107 g/L DCW) and rhIFN-_ concentration (26.5 g/L) after 16.5 h were achieved. For Nonlinear identification of recombinant E coli Fed-batch fermentation Wiener model based on neural network were used. The linear part of Wiener model is represented in state space form and nonlinear part is estimated with a single layer neural network. After identification Wiener Model is optimized with numerical optimization algorithm. Validation results show good compatibility with experimental data and Euclidean norm of the estimated outputs error are less than 2.72 and residual errors between estimated outputs and experimental data are less than 0.01.

Index Terms E. coli, Fed-batch fermentation, MOESP algorithms, neural network, numerical optimization, Wiener model.