Soft sensor modeling method for penicillin fermentation process based on SPA-WOA-SVR
A soft senor method based on SPA-WOA-SVR is proposed to address the engineering technical challenge of difficult direct online measurement of key parameters during penicillin fermentation process.The Continuous Projections Algorithm(SPA)is used to select auxiliary variables for the fermentation process.The selected feature auxiliary variables(CO2 concentration,dissolved oxygen concentration,pH value,acid flow acceleration rate,alkali flow acceleration rate)are used as inputs,and the key parameters of the fermentation process(bacterial concentration,substrate concentration,and product concentration)are used as outputs.A soft senor model is established based on Support Vector Regression(SVR).To improve the prediction accuracy and stability of the soft senor model,Whale Optimization Algorithm(WOA)is used to optimize the important parameters of the model(kernel width,penalty factor,and insensitivity coefficient).The constructed soft senor model was applied to predict key parameters in the penicillin fermentation process.Simulation results showed that,compared with the traditional SVR soft senor modeling method,using the SPA-WOA-SVR soft senor method,the determination coefficients of key parameters and test sets were all above 0.953 3,and the root mean square error was less than 0.029 3.Especially,the bacterial concentration increased from 0.904 4 to 0.987 9,and decreased from 0.031 4 to 0.009 8.This indicates that the SPA-WOA-SVR based soft senor modeling method effectively improves the predictive performance of the model,has higher prediction accuracy and stability,and can be applied to soft senor modeling of general nonlinear systems.
国家科技期刊平台
NSTL
万方数据
