Qualitative Identification And Quantitative Analysis Of Edible Vegetable Oil Based On Raman Spectroseopy

As a fast, efficient, non-destructive optical technique, Raman spectroscopy has been successfully applied in diverse industrial fields in recent years. The quality, doping detection and quantitative analysis of edible vegetable oil is a hot research topic in food field. Their traditional chemical analysis has the problem of long experimental time and resulting environment pollution, so there is an urgent need for quick, accurate detection methods. In this paper, Raman spectroscopy is introduced to analyze the qualitative identification and physicochemical properties of edible vegetable oil. The methods of the edible vegetable oil variety identification, olive oil doping and the determination of Grease iodine and saponification value are proposed. The main contents of this thesis are as follows:1. Summarize the principle, features, application and future development direction of Raman spectral analysis technology. Introduce Raman spectra preprocessing algorithms. Systematically elaborate the typical calibration methods and outline the applications of Raman spectroscopy in edibie oil quality control.2. Propose a novel method to fast discriminate edible vegetable oils by Raman spectroscopy. To identify the unknown samples automatically, we build a library of training samples based on known samples. Based on their original Raman spectra, baseline correction and normalization are applied to obtain standard spectra. Two characteristic peaks describing the unsaturated degree of vegetable oil are selected as feature vectors; then the centers of all classes are calculated. For an edible vegetable oil with unknown class, the same pretreatment and feature extraction methods are used. The Euclidian distances between the feature vector of the unknown sample and the center of each class are calculated, and the class of the unknown sample is finally determined by the minimum distance. Experimental results show that, the above classification model can be applied to discriminate unknown edible vegetable oils rapidly and accurately.3. For there is still lack of a rapid and simple detection method for doping identification of olive oil. A fast discrimination method based on Raman spectra using least squares support vector machine was presented. Firstly, some known class olive oil samples were chosen randomly as training samples and their original Raman spectra were obtained, then a pretreatment and band selection were made for those spectra, and then, the LSSVM classifier was built. Secondly, for the raman spectra of unknown test samples, the same pretreatment and band selection were used. Finally, the discrimination results were attained through the ESSVM classifier. Experimental results show that, this method was able to discriminate olive oil adulteration and the lowest detection limit of the doping amount was5%. Compared with other classification methods, LS-SVM classifier has the best classification performance.4. In order to solve the problem of the poor reproducibility and vulnerable to grease itself color interference when detecting the saponification and iodine value of the grease, a new method based on Raman spectroscopy is presented. There is a strong linear correlation between the saponification value, iodine value and there Raman spectra. So quantitative analysis model is established by partial least squares method. The reproducibility and repeatability of the model meet the requirements of National Institute of Standards.