Research On Online Optimization Of Penicillin Fed-Batch Fermentation Process

Microbiologic fermentation is an important part of biological engineering,which has been widely used in medicine,food,chemical and many other fields.Penicillin can be obtained by fermentation as a kind of antibiotics.Because of its significant effect,the production and application have attracted widespread attention.However,penicillin fermentation process currently has low yields and high production costs and optimal control is an effective way to solve such problems.Current research on penicillin fermentation process optimization is mainly based on the process model and uses offline optimization method.However,the fed-batch fermentation process is non-linear,random and uncertain.During the production process,the problem of model mismatch often occurs and the offline optimization results are difficult to implement.In order to improve the production quality and economic benefits,the online optimization is of great practical values for penicillin fermentation process.Therefore,taking penicillin fed-batch fermentation process as the object,this thesis comprehensively and systematically studied its modeling and online optimization problems.The main contents are summarized as follows:1.Based on the detailed analysis of the reaction mechanism of penicillin fed-batch fermentation,a simplified process mechanism model is established with the bacterial cell concentration,substrate concentration,product concentration and fermentation broth volume as the state variables.Differential evolution algorithm is used to identify the unknown parameters of the model.The simulation results show that the mechanism model can describe the trend of the key variables during the fermentation process,but the model accuracy is not reliable.2.Aiming at the problem that the mechanism model is not accurate enough,a model updating strategy is proposed.When the deviation between the predicted value of the process model and the actual value exceeds the set threshold,just-in-time learning algorithm is used to find out the data similar to the current working point from the historical fermentation data to construct the learning set.And the Least Squares Support Vector Machine is used to establish the local model,so that the penicillin fed-batch fermentation process model is updated online.The simulation results show that the model updating strategy adopted ensures the model accuracy and real-time modeling.3.Aiming at the characteristics of penicillin fed-batch fermentation process,an online segmentation optimization method is proposed.The fermentation process is divided into two stages:growth stage and production stage.Then,optimization models for the two stages are established,respectively.Fuzzy clustering algorithm is applied to determine the current stage of the fermentation process,and the corresponding optimization model is then chosen.An improved particle swarm optimization algorithm is used to solve the optimization problem,and the simulation results show the effectiveness of the optimization method.Finally,the main work of the thesis is summarized,the further research direction on modeling and optimization of penicillin fed-batch fermentation process is prospected,and the problems to be studied in the next stage are discussed.