Statistical Modelling of Glutamate Fermentation Process Based on GAMs
Application of Generalized Additive Models (GAMs) for modelling of Glutamate fermentation process was proposed in this paper. There were so many variables in fermentation process and insignificant variables that might worsen pre-built model performance, so experiments of choosing significant variables were firstly carried out. One new model was constructed after choosing time (Time), dissolved oxygen (DO) and oxygen uptake rate (OUR) as significant variables. The simplified relationships that could reflect each variable effect in fermentation process between Time, DO, OUR and GACD were investigated using the constructed model. The integrated relationships that could provide theoretical base to implement control and optimize in fermentation processes between Glutamate and other significant variables were also explored. Normally, fermentation model was specific with the character of poor generalization, because of the complications of fermentation process, high degree of time-varying and batch changing. However the new model fitting results indicated the advantages, in term of non-parameter identification, prediction accuracy and robust ability. So the new model in this paper was satisfiedly characteristic of generalization. The advocated modelling method potentially supplies an alternative way for optimization and control of fermentation process.
Glutamate Fermentation Statistical model GAMs
Chunbo Liu Xuan Ju Feng Pan
Institute of Automation, Jiangnan University, Wuxi, 214122, China
国际会议
无锡
英文
490-499
2010-09-17（万方平台首次上网日期，不代表论文的发表时间）