Research Of Identification And Adulteration Of Edible Oil By Near-infrared Spectroscopy

Edible oil is an important part of people's dietary pattern, and its qualities are directly related to human health. In order to protect the benefits of the legitimate producers and consumers, it is necessary to establish a simple, rapid and effective method for detecting the quality and adulteration of the edible oils. Near-infrared spectroscopy(NIRS), characterized by fast, convenient and non-destructive, especially is applicable to some relatively high viscosity liquids(such as edible oils). By using near-infrared spectroscopy combined with chemometrics methods, models for categories identification of edible oils, qualitative identification and quantitative detection of the different adulterated systems of peanut oil and sesame oil was built in this paper.1. In the categories identification of the five kinds of pure edible oils, 102 samples of peanut oil, sesame oil, soybean oil, cottonseed oil and palm oil were scanned by NIR spectrometer. In the selection of the suitable wavelength and spectral pretreatment, Clustering Analysis(CA) and Discriminate Analysis based on Principal Component Analysis(DA-PCA) were adopt to set up a model of identification of edible oil samples. The result showed that the two analysis methods were able to distinguish the five kinds of edible oils. The model established by the DA-PCA to verify the pure edible oils of the recognition rate was up to 100%, which is better than using CA.2. Qualitative and quantitative determination of the adulterated binary system of pure peanut oil and sesame oil samples which was respectively mixed with 0~100% concentrations of other edible oils were researched. The adulterated oil and the pure oil samples were verified by the qualitative analysis models set up by the discriminate analysis. In addition, the quantitative analysis models was built respectively by Principal Components Regression(PCR), Partial Least Squares(PLS) and Modified Partial Least Squares(MPLS) methods, and the optimal calibration models was validated by the validation sets. The main research results are as follows:(1) In the qualitative analysis, the correct classification rate of the peanut oil mixed respectively with the soybean oil, cottonseed oil and palm oil was 99.5%, 90.7% and 96.0%, respectively. The lowest limit in the detection of the peanut oil mixed with cottonseed oil was 5%. The correct classification rate of the sesame oil mixed respectively withed the soybean oil, palm oil reached 96.8% and 94.9%, respectively.(2) In the quantitative analysis, comparing with PCR and PLS, all the optimal models of the peanut oil and sesame oil adulteration were obtained by MPLS calibration method. Using validation sets to validate the optimal models of the peanut oil mixed respectively with the soybean oil, cottonseed oil and palm oil, the result was that the determination coefficient of prediction(R2p) was 0.998, 0.997, 0.995; the relative standard deviation(RSD) was 2.327%, 3.040%, 3.830%; and the relative prediction deviation(RPD) was 3.542, 2.642, 2.581, respectively. Using validation sets to validate the optimal models of the sesame oil mixed respectively with the soybean oil and palm oil, the result was that R2 p was 0.988,0.985; RSD was 3.504%, 4.965%; RPD was2.969,2.409, respectively. These results showed that the optimal calibration models of the adulterated binary system of the peanuts oil and sesame oil adulteration were able to be used for the quantitative detection, and the prediction accuracy of the model established by the peanut oil mixed with soybean oil was highest; the lowest limit in the detection of the peanut oil mixed respectively with cottonseed oil and palm oil were 3% and 5%.3. Qualitative and quantitative determination of the adulterated ternary system of pure peanut oil and sesame oil samples which was respectively mixed with 0~100% concentrations of other two kinds of edible oils were researched. The main research results are as follows:(1) In the qualitative analysis, the correct classification rate of the peanuts oil and sesame oil adulteration reached 94.5%, 92.4% respectively, and the model of the sesame oil adulteration had lost the ability to identification when the adulteration quantity was below 3%.(2) In the quantitative analysis, comparing with PCR and PLS, all the optimal models of the peanut oil and sesame oil adulteration were obtained by MPLS calibration method. Using validation sets to validate the optimal models of the peanut oil together with the soybean oil and palm oil, the result was that R2 p was 0.992 and 0.997; RSD was 2.748% and 2.086%; RPD was 2.805 and 3.487, respectively. While using validation sets to validate the optimal models of the sesame oil together with the peanut oil and soybean oil, the result was that R2 p was 0.985 and 0.982; RSD was 4.677% and 5.088%; RPD was 2.737 and 2.573, respectively. These rerults showed that the optimal calibration models of the adulterated peanut oil ternary system would obtain good prediction in the practical detection, but the prediction accuracy of the optimal calibration models of the adulterated sesame oil ternary system was not ideal.