Multi-objective Optimization Control Of Escherichia Coli Fermentation Process

Escherichia coli is widely used in genetic engineering as a recipient of exogenous genes due to its genetic stability and simple culture.It is of great economic significance to optimize the fermentation process of Escherichia coli and improve the fermentation efficiency and production of Escherichia coli.However,due to the non-linearity,time-variability,uncertainty in the fermentation process,and the inhibition of by-product accumulation on cell growth,the traditional control method is difficult to achieve adequate control of the fermentation process.Given the above problems,according to the fermentation characteristics and actual industrial conditions,the optimal control of the Escherichia coli fermentation process is realized by optimizing the feeding trajectory.The idea of receding horizon optimization and feedback correction of model predictive control is applied to the fermentation process.The optimal control method is used to solve the predictive control problem,and the parameter identification is added into the control scheme to enhance the robustness of the control system.The main contents of this paper are as follows.Firstly,the fermentation kinetic model and control objectives are simplified.The control objectives are set to improve the concentration of bacteria,reduce the accumulation of acetate and smooth the feeding process.However,in the actual fermentation process,the key parameters such as bacteria concentration are not easy to be directly measured,so the fermentation kinetic model and control objectives are simplified.The parameters that are not easy to be directly measured are represented by the parameters that can be directly measured or calculated,which reduces the detection conditions and facilitates the control operations.Based on the simplified control objective and kinetic model,the fermentation process is optimized.Taking advantage of the characteristics of model predictive control in dealing with nonlinear problems and constraints,this paper combines predictive control with intelligent algorithms,using particle swarm optimization algorithm with constriction coefficients and multi-objective particle swarm optimization algorithm combining with TOPSIS(Technique for Order Preference by Similarity to an Ideal Solution)method to solve the receding horizon optimization problem of model predictive control.The simulation results are compared with other literatures.The simulation results show that the two schemes have significant effects on improving the concentration of bacteria and controlling the production of acetate.Even if the initial and external conditions are the same between different fermentation batches,the kinetic parameters between different batches will also be different due to bacterial activity and other unknown factors.Robust model predictive control is used to calibrate the model online and optimize the fermentation process according to the mismatch of the kinetic model.The model parameters are identified.Particle swarm optimization algorithm is used to identify and update the model parameters online at each sampling time.Theil's inequality coefficient(TIC)is used to test the consistency of the simulation results of the above model adaptive correction scheme.The analysis results show that the above scheme has good tracking performance for the system state.Several groups of random parameter values are generated by Monte Carlo method and substituted into the kinetic model as the real model parameter values.The simulation experiments are carried out on the above models,and the simulation results prove that the control system has good robustness under the condition of model mismatch.