Study On Near-infrared Spectroscopy Analysis Methods Of Multi-component Organic Compounds And Petroleum Products

Near-infrared spectroscopy analysis has been widely used in various fields as a rapid analysis method.But a universally recognized method hasn’t been found about the quantitative analysis of multi-component mixed samples in near-infrared spectroscopy analysis,and the actual samples are mostly mixture.Therefore,the quantitative analysis of multi-component mixture is very necessary.This thesis mainly studied the method for rapid quantitative analysis of multi-component complex system.It was established via combining the multivariate calibration methods with near-infrared spectroscopy analysis.The experiment included as follows:In the first chapter,the development and application of near-infrared spectroscopy analysis technology were reviewed in recent years,In addition,and several multivariate calibration methods were sketched about near-infrared spectroscopy.In the second chapter,the applications of different multivariate calibration methods were researched for determining protein content of soybean.The model of quantitative relationship was built based on the near-infrared spectral data of protein content in soybean by using of partial least square(PLS)and principal component regression(PCR).The prediction performance of the establish models is assessed by external test validation and leave one out cross validation.It is demonstrated that the built model of two different multivariate calibration methods were feasible in quantitatively determining protein content.Compared with PCR,the prediction error of the model established by PLS is smaller and the correlation is better.In the third chapter,The PLS and PCR respectively combined with near-infrared spectroscopy to determining the oil content of soybean.The result shows that the PCR method has higher accuracy.In the fourth chapter,two effective substances in Zuguangsan were simultaneously quantitatively analyzed by near-infrared spectroscopy compared with PLS.The obtained model was subsequently validated by determining the active substance of a series sample of test set.The result demonstrates that the established analytical model is suitable for nondestructive and fast analysis of Zuguangsan.The method is effective and can be satisfied with the requirements of determining the effective substance.In the fifth chapter,the PLS method was used to construct the analysis model for cetane number of original near-infrared spectroscopy and first derivative spectroscopy approach based on the ASTM D–613 method and near-infrared spectroscopy of diesel oil.The result shows that it can establishes optimum model and obtain more accurate predicted result by spectra pretreatment,the accuracy of prediction can be improved.