Abstrato

Study of glutamic acid fermentation process hybrid modeling

Guicheng Wang, Jinjin Ma, Changliang Guan, Jinna Li

Fermentation process of glutamic acid is a very complicated industry process, which needs different environment parameters and has seriously nonlinear and instability. Studying the kinetic model and neural network model combining hybrid modeling of glutamic acid fermentation process, hybrid model consists of two parts, one is the mechanism model, through reasonable assumptions and limitations, simplifying the kinetics of biochemical reaction process, it is basis of hybrid model, describes the basic characteristics of glutamic acid fermentation process; the other part is a neural network model, using neural network training function of knowledge, and establishing glutamic acid fermentation process between the input and output mapping, this mapping only depends on the actual production data, and has nothing to do with the actual process, it is a secondary part of the hybrid model for correction kinetic model. Through simulation, respectively, glutamic acid fermentation process kinetic model and hybrid model to compare the results to prove the value of hybrid model predictions closer to the system actual output, the model is more accurate.