Study On Oil Detection Method Based On Raman Spectroscopy

Raman spectroscopy technology is applied to analyze the rapid detection of oil. Providing test data for the oil testing, grading and evaluating is of great significance for oil quality and to maximize the economic and rational use. Because of its huge economic benefits, oil detection analysis technique has been the focus of academic research. There are time-consuming, expensive and complicated shortcomings in the traditional chromatographic methods, it is difficult to achieve on-site rapid testing. In this paper, the Raman-based detection method of edible vegetable oil and gasoline, combining with support vector machine and partial least square is studied, the main research contents and results are as follows:1. Rapid oil detection method research and analyzer system design. The rapid detection method of oil is researched according to the Raman spectroscopy of edible vegetable oil and gasoline, and a portable oil quality analyzer is designed with these research results. This analyzer can predict fatty acid contents in edible vegetable oil and gasoline contents. The time of testing a sample is within1-2minutes, and the accuracy rate exceeds90%, the theoretical study is carried to practical use successfully.2. Fatty acid content detection of edible vegetable oils.Raman spectroscopy combined with multi-output least square support vector regression(MLS-SVR) and least squares regression(PLS) is applied to establish a quantitative analysis model to predict fatty acid content in edible vegetable oil, it is able to predict the three fatty acid content: saturated fatty acid(SFA), oleic acid, linoleic acid. Results show that, MLS-SVR is better than PLS, the RMSEP of the three fatty acid are0.4967%,0.8400%,1.0199%, the correlation coefficient(R) are0.8133,0.9992,0.9981; and the RMSEP of unknown sample are1.9171%,3.1932%,4.8149%, the R are0.6133,0.9771,0.9802, the prediction error does not exceed5%.3. Gasoline contents detection.The experiments of doping93#,97#gasoline with benzene, aromatics, olefins, methanol, ethanol are conducted using the rapid oil analyzer system. First the correspondence between Raman peaks and gasoline contents is researched, then the relevant peaks are extracted as eigenvalues, The MLS-SVR and PLS models are established to analyze the quality of gasoline. Results show that, MLS-SVR is better, its RMSEP of the five key quality indicators of gasoline are0.22%,0.27%,0.27%,0.17%,0.14%, the R are0.9986,0.9993,0.9985,0.9923,0.9935, the prediction of unknown sample also has a good accuracy.This research provides a simple, accurate, and non-destructive method for the rapid detection of oil contents.