Research Of Spectrum Rapid Identification Methods Of Edible Oil Species And Quality

Edible oil security has become a common topic of concern to many people.But because of the lack of an accurate,simple and quick detection technique of edible oil,edible oil quality and safety issues are still showing a state of repeated prohibitions.In reviewing the traditional effective detection technique we can find,its equipment is not only expensive,but also the operation and maintenance is very complex,and cannot be widely applied to the field of law enforcement inspection.Therefore,establish an accurate,simple and rapid method to detect the authenticity of edible oil and provide technical support for the field law enforcement,is benefit to accelerate the development of national standards for edible oil,to prevent the occurrence of edible oil quality and safety issues,and safeguard the interests of consumers and legitimate production and marketing companies.Near infrared spectroscopy is the middle of last century developed a detection technique,using the spectrometric method Analyze composition and structure of the substance,it has rapid,non-destructive,low-cost,no complex pre-processing,environmental protection and many other features.This paper is research on using the Combination of near-infrared spectroscopy and support vector machine,in view of the characteristic index of edible oil in the national standard,the method of combining quantitative and qualitative analysis is adopted to realize the detection and identification of edible oil.The main research contents and conclusions are as follows:(1)Established a peroxide value quantitative model of edible oil.Gathered a certain number of representative edible oil samples and collected its near-infrared spectrum.The peroxide value of the sample was determined by the iodine value method,and the result was used as the input value of the quantitative model of peroxide value.Used support vector regression(SVR)establish a quantitative models of peroxide value,and optimized selection spectral pretreatment method and parameter optimization method,and finally establish the best quantitative model of peroxide value.The results show that,for the quantitative model of peroxide value,the correlation coefficient of the Calibration set is 99.9999%,and the correlation coefficient of the prediction set is 90.139%.It means that the model can accurately predict the peroxide value of edible oil,and accurately determine the quality of edible oil.(2)Established a quantitative model of edible oil fatty acids content.Gathered a certain number of representative edible oil samples and collected its near-infrared spectrum.The palm acid,oleic acid and linoleic content acid were determined by gas chromatography,and the results were as the input values of the three kinds of fatty acid content quantitative models respectively.Used support vector machine regression(SVR)establish three kinds of fatty acid content quantitative model respectively,and optimized selection parameter optimization method,and finally establish the best three kinds of fatty acid content quantitative models.The results show that,for this three kinds of fatty acid content quantitative models,the correlation coefficient of the prediction set were 95.0876%?99.8592% and 98.5951% respectively.It means that the models can accurately predict edible oil fatty acids content.(3)Established a quantitative-qualitative classification model of edible oil which judged by fatty acid content.Used support vector machine classification(SVC)combined with three kinds of fatty acid content real data,established a qualitative model of edible oil,and verified the feasibility that the types of edible oil judged by fatty acid content.Combined use of the fatty acid content quantitative model and the edible oil qualitative model,established a quantitative-qualitative classification model of edible oil which judged by fatty acid content.The results show that,the accuracy rate of the model is 100%,can identify the types of edible oil quickly.