Optimization Of The Fermentation Process Based On Its Hybrid Model

Fermentation is a process that produces and accumulates the desired metabolites by chemical and physiological changes during microbial growth and culturing. With the improvement of the fermentation industry, optimal control of fermentation processes is becoming increasingly important. Nosiheptide is a sulfur peptide antibiotic that can be produced by fermentation. When used as feed additives, it can significantly promote animal growth and leave no residue in animals. Nosiheptide has a broad market prospect, but the production is low currently. Optimal control is an important way to solve this problem. Taking nosiheptide fermentation process as the background, modeling and optimization of nosiheptide fermentation process is considered in this thesis, whose main work is summarized as follows:1). Based on analysis of the technical flow and main factors of nosiheptide fermentation, mechanism model of nosiheptide batch fermentation is studied according to fermentation kinetics and material balance principles. The unknown parameters in the model are identified using particle swarm optimzation. Mechanism model of nosiheptide fed-batch fermentation, which lays a solid foundation for the further development of nosiheptide fermentation hybrid model, is then developed based on mechanism model of nosiheptide batch fermentation.2). Due to complexity of fermentation process, the mechanism model is developed after some simplifications and assumptions for the process, so it is not precise enough. So, a hybrid modeling method that combines the mechasim model and the empirical model is developed. The mechanism model is used to describe the basic knowledge, and RBF neural network is used to compensate for the error between the mechanism model and the practical process. Thus the accuracy and generalization ability of the hybrid model are improved effectively.3). The standard particle swarm optimization is easy to fall into local optimum, so a chaotic migration operator is introduced into PSO and an improved PSO-chaotic particle swarm optimization (CPSO) is proposed. Simulation results of typical test functions show that the algorithm has good performance and it can be used to solve complex optimization problems. Aiming to solve the problem that nosiheptide fermentation process has low production currently, an optimization model with the objective of maximizing the production is established. Then decision variables and constraint conditions are decided. Using the developed hybrid model to describe nosiheptide fed-batch fermentation, the optimization problem is solved and analyzed with CPSO.