Industrial Processes Based On Ls-svm Soft Sensor Modeling Method Study

Soft-sensor technology is a kind of valid method to solve complex measuring tasks and implement the on-line evaluation of the variables in industrial process.In recent years, Soft-sensor technology was widely used in the biological and chemical areas.Microbial fermentation is a complex and non-linear biochemical process,the accurate measurement of biomass concentration is the premise to improve the fermentation yield and quality,so as to achieve optimal control.To address this measurement problem,a soft-sensor model based on least squares support vector machine(LS-SVM) of fermentation process was proposed.This thesis focuses on soft sensor modeling method based on LS-SVM,the main researches of the thesis boils down to the following aspects:1.For the problem of online measurement of GSH concentrations,a soft-sensor model based on LS-SVM was established.The feasibility of the model was proved by simulation results.The mechanism of fermentation,auxiliary variable selection and sample data processing were studied.2.To resolve the disadvantages of generalization ability of soft-sensor model based on LS-SVM,the wavelet transform was added to improve the model.The soft-sensor model based on W-LSSVM was used to measure GSH concentration.3.Model parameter is a key factor which affects the model performance,so a parameter selection method based on genetic algorithm(GA) was studied to select LS-SVM parameters.Due to the inbreeding problem,parameter selection method based on GA was improved.Compared the experimental results between the two genetic algorithms,the improved GA gave the better performance.The thesis proposed a soft-sensor model based on improved GA-WLSSVM to resolve the problem of online prediction of GSH concentration,and the simulation results showed that the proposed model is preciser.