Raman Spectral Synthesis And Its Application In The Analysis Of Methanol Gasoline

Raman spectroscopy is a kind of molecule scattering spectroscopy, which can analyze the structure, components, concentrations and some other properties of samples. It is being used in many areas, including chemistry, physics, biology, medicine, etc. because of its non-invasion, fast response speed and high reproducibility. However, as a kind of indirect analysis method, the precision of the Raman analysis model is greatly dependent on the quantity, distribution and accuracy of training data, and it is difficult to collect complete training data in a practical application. In this thesis, Raman spectral synthesis is proposed to break through the above limitations. The main content are included as follows:①Summarize the principle, characteristics and applications of Raman spectroscopy. Systematically elaborate the common spectral pretreatment and typical calibration methods. Present the limitations of Raman spectroscopy and its disadvantages in quantitative modeling.②A Raman spectral synthesis model is proposed to overcome the difficulty in collecting complete training data. The Raman spectra of seven common-used gasoline additives, which have great influence on gasoline quality, are measured and pretreated. Based on the linear relationship between the feature peak intensity of an additive in gasoline Raman spectra and its concentration, the corresponding spectral synthesis model as well as parameter identification method are put forward. The synthesis model is proved by comparing the measured spectra and the synthetic spectra.③Based on the above spectral synthesis model, the determination of methanol ratio in methanol gasoline is studied. The Raman spectra of12methanol gasoline samples are measured and pretreated. With the traditional PLS (partial least square) calibration algorithm, the SECV (standard error of cross validation) is1.463%. However, the SEP (standard error of prediction) increases to4.248%when only4training samples are used. To improve prediction precision, the Raman spectra of the remaining samples are composed with the above spectral synthesis method. The SEP of improved model decreases to1.512%. ④Based on the above research result, a portable fast analyzer for methanol gasoline is developed, which can determine seven key quality indexes such as methanol ratio etc. The analyzer is introduced from the aspects of instrument appearance, user interface, hardware and software. With fast analysis speed, high analysis efficiency, less sample consumption and less training data, this analyzer has broad application prospects in the methanol gasoline field.