Research On Adaptive Fuzzy Control For The Fermentation Process

Fermentation process possesses non-linear and time difference, its.inherence mechanism is very complex, some important course variables can't measure online, which made the fermentation process modeling and controlling more complex. Aiming at the research of the glutamic acid fermentation process, intelligent control theory is used on the study of the modeling and control methods of the fermentation process.Based on a'great deal of field recording data in monosodium glutamate process, a multi-variable neural network dynamic model of the fermentation process is established. Then we combine the advantages of fuzzy control technology, neural network and genetic algorithm to set up a adaptive fuzzy neural network controller based on genetic algorithm. At first, we adopt genetic algorithm to tune the parameters of antecedent variables'membership functions and the output values; then we apply BP algorithm to tune these parameters again online to optimize controller. Simulation is conducted based on the fermentation process established by the neural network model. The simulation also takes into account the existence of the cell concentration measurement areas of interference. The results show that the controller of glutamic acid fermentation process can be effectively controlled, and can effectively adapt to changes of the operation and environment.Finally, combining the offline most optimal control method and online adaptive fuzzy neural network control method, the hybrid fuzzy adaptive fermentation process control method is designed. The offline optimization track is the main control. And the output of adaptive fuzzy controller is the auxiliary control, in order to amendment the main control online. The simulation results show that application of hybrid fuzzy adaptive controller can not only guarantee a higher rate of acid production, but also effectively prevent the fermentation process in a series of interference.