Designing new controller for fed-batch cultivation of recombinant E.coli

Fed-batch fermentation processes are common methods of producing biologic recombinant cells from different microorganisms. Model-based control of bioprocesses is a difficult task due to the challenges associated with bioprocess modelling and the lack of on-line measurements. So the designed controller should be robust against uncertainties and parameter variations. This paper surveys a feeding control approach in fed-batch cultivation of high cell density E.coli producing recombinant proteins based on neural networks. For this purpose, supposing that key variables are not available by sensors, first an observer is introduced to extract necessary variables for the control strategy. Then a neural network self-tuning PI controller is designed and some advantages and disadvantages of this controller are denoted. It is shown that the controller is capable to reject disturbances and tolerate parameter variations.