Research Of The Substrate Feeding Control Strategy In Fermentation Process

In Fed-batch fermentation process, the substrate feeding control relates to the production of fermentation, this is the key problem in the control of the fermentation process. At present, the control of the amount usually depends on many experiences, the reliability is not good. The optimization of the speed of material makeup if general though the study the fermentation's mechanism model, the optimization calculation is based on fermentation process's measurable physical parameters, the effect isn't good because the fermentation process is time-varying. Through the study of modeling and optimization the fermentation process, which have great significance in improving the fermentation production.This article takes the glutamate fermentation as the research object, researching the modeling in glutamate fermentation and substrate feeding's optimization.In accordance with the features of non-liner, time varying and uncertainty for the fermentation process, a modeling method bases on the principle of Structure Risk Minimization such as the Support Vector Machine (SVM) method is improved. Choosing the kernel function is a key question in support vector machine's theory, through the study of kernel function, to combine the local characteristic of RBF kernel function with the overall characteristic of Polynomial kernel function. A Mixture Kernel Support Vector Machine (MKSVM) forecast model for glutamate fermentation is established. The simulation results show that the model's learning ability and generalization ability have reached good effect.On the basis of glutamate forecast model, taking the fermentation production as a goal to optimize the speed of substrate feeding. Through compare the Particle Swarm Optimization (PSO) and the Quantum-behaved Particle Swarm Optimization (QPSO), using the QPSO to optimize the substrate feeding in fermentation process. The main control is fed rate of substrate sugar, to seek the optimal curve of material makeup rate and control it, making the production most at the end of the fermentation. The glutamate fermentation experiment's results show that using the methods to optimize the substrate feeding, the fermentation production of glutamate has been improved.With the study of modeling and substrate feeding optimization of glutamate fermentation process, using the advantage of the interface of Visual Basic and powerful simulation of Matlab, taking advantage of the method about mixed programming, a software of soft measurement and optimization for glutamate fermentation is designed.