Rapid Determination Of Anaerobic Degradation Of Algaes And Other Plants By Near Infrared Spectroscopy

Energy algaes and aquatic plants and energy account for a great proportion of organic waste.Anaerobic degradation of energy algaes and aquatic plants is one of the most promising technology in utilization of organic wastes,and its research and utilization have been widely carried out.Anaerobic biodegradatiligy is an important index in the process of anaerobic fermentation.Usually,biochemical methane potential(BMP)is used to represent the anaerobic degradation of the material.Direct measurement method and component analysis method are the traditional measuring methods of BMP,but these methods are expensive and time-consuming.This paper puts forward the method to measure the BMP of aquatic plants and energy algaes by near infrared spectroscopy combined with chemometrics methods.The raw data of samples BMP were obtained from the experimental platform.At the same time,near infrared spectral data are obtain by Fourier transform near infrared spectrometer.Establish a reliable mathematical model to associate raw BMP data with the spectral data so as to realize the rapid prediction of aquatic plants and energy algaes BMP through PCA,PLS and r PLS in the full spectrum.In order to improve the prediction accuracy of the model,a method of band feature band selection is proposed.SPLS improves the accuracy of mathematical model and ability of explaining,but the filtered data can be further reduced,thus a new genetic algorithm with support vector machine learning method is proposed.This method has good global search capability and eliminate the redundant information of samples,it is suitable for small samples.According to the results of model evaluation,it is known that the accuracy and stability of the prediction model based on GA-SVM is good.Compared with PLS,RPLS and SPLS,the RMSEP is reduced by 46.27 m L,33.82 m L,10.19 m L,R2 is increased by 8.23%,7.26%,3.1%.In order to ensure the applicability and generality of the model,the research work of the model transfer is also carried out.The main research content is divided into the following three aspects:(1)Combined with existing experimental data in the literature,analyse the correlation between process parameters and BMP values and find the maximum impact factors as the main monitoring objects in the BMP test,in order to reduce the error of the data.(2)The near infrared spectra of the samples were collected by Fourier near infraredspectroscopy,and the effects of different algorithms of orthogonal signal correction are studied and evaluated by band assignment of NIR.Establish a reliable mathematical model to associate raw BMP data with the spectral data so as to realize the rapid prediction of aquatic plants and energy algaes BMP through PCA,PLS and RPLS in the full spectrum;model transfer is also carried out.(3)In order to further improve the prediction accuracy of the model,the simplified modeling data,bands selection of spectral data is carried through SPLS and GA,and combined with the band assignment to analyze whether the characteristic bands are representative after screening;the support vector machine regression model is established to analyze the effect of the model.