Study On Techniques Of Authenticity Spectral Detection Of Walnut Oil

The standard of living of people has steadily increased,and the demand for edible oil by the people has also increased.Some lawless elements opportunistically use low-quality,low-cost edible oils to be blended into high-quality edible oils,and the fish are mixed with the eyes to make a profit.In view of the mixed market of edible oil,the traditional physical and chemical detection methods not only require the operation of personnel with professional knowledge,but also the detection time is long and the detection process is complicated.Therefore,it is urgent to explore a rapid detection and identification technology method that can be used for the quality of edible oil.Based on the laser near-infrared spectroscopy technique and walnut oil as the research object,the research on the identification of walnut oil adulteration was carried out.The binary qualitative identification and quantitative prediction model of walnut oil was established,and the ternary admixture of walnut oil was established.Identification and quantitative prediction model,quantitative prediction model of linoleic acid content of walnut oil.The main contents are as follows:Research and explore the technical method of binary doping qualitative identification and quantitative prediction of walnut oil.and establish a binary qualitative doping identification and quantitative prediction model of walnut oil.Two preprocessing methods,multi-scattering correction algorithm(MSC)and standard normal transformation and de-trend combination(SNV-DT),combined with continuous projection algorithm(SPA)and competitive adaptive method and partial least squares method(CARS-PLS)two feature variable extraction methods,through genetic algorithm(GA)and cross-validation(CV)two parameter optimization algorithms,using support vector machine(SVM)to establish a variety of walnut oil binary adulteration Qualitative identification(SVC)and quantitative prediction(SVR)models were used to study the modeling techniques and methods for finding the qualitative identification and quantitative prediction model of the optimal pseudo-authenticity of walnut oil.The experimental results show that the MSC-SPA-SVC model is the best predictive performance of the binary-adapted pseudo-authenticity of walnut oil.It can realize the binary identification of walnut oil.The binary quantitative adulteration of walnut oil is predicted by MSC-CARS-PLS.The correlation coefficient R of the prediction set of SVR model is as high as 97.1715%,which indicates that the model can better realize the binary quantitative pseudo-predictive function of walnut oil.Research and explore the technical method of ternary doping qualitative identification and quantitative prediction of walnut oil,and establish a ternary doping qualitative identification and quantitative prediction model of walnut oil.The SNV-DT and MSC preprocessing methods,SPA.CARS-PLS and joint interval partial least squares(SiPLS)three feature variable extraction methods,GA,PSO and cross-validation K-CV three parameters optimization algorithm Combined with support vector machine,a variety of pseudo-qualification identification and quantitative prediction models for walnut oil ternary doping were established,and the modeling techniques and methods for determining the optimal identification and quantitative prediction model of ternary adult pseudo-authenticity were studied.The experimental results show that the accurate rate of SNV-DT-SPA-SVC model for walnut oil ternary adulteration is up to 99.6667%,and the prediction of MSC-SPA-PSO-SVR model is the bestgeneralization ability.Study and explore the quantitative prediction method of linoleic acid in the characteristic components of walnut oil.Two pre-processing methods,MSC and SNV-DT,combined with SPA,CARS-PLS and BiPLS,three different characteristic wavelength extraction algorithms,and three optimization parameters of GA,PSO and K-CV are used to find ten kinds of walnuts.Oil characteristic component linoleic acid quantitative prediction model.The experimental results show that the data processing methods are different,which leads to the difference in model prediction performance.The correlation coefficient R of the calibration set of MSC-SPA-GA-SVR model is 95.7665%,and the correlation coefficient R of the prediction set is 96.7219%.above 95.By analyzing and comparing the differences in data processing methods of ten model modeling,the optimal model for quantitative prediction of linoleic acid in walnut oil was found.