Application Of Near Infrared Spectroscopy In Several Foods

Near-infrared spectroscopy has wide application in food detection.In this paper,two typical food forms—liquid and solid,including apple juice,hemp oil,lotus seed powder,Tetrastigma Hemsleyanum and Panax pseudoginseng,were selected as representative targets.Near-infrared spectroscopy combined with chemometrics were employed for rapidly and accurately non-destructive detection for food components analysis,classification,and quality identification.Those results provide a new and rapid detection method to promot food safety and safeguard the consumer rights.The main results were as follows:(1)Based on the near-infrared spectroscopy of apple juice,the partial least-squares(PLS)regression model was established.After normalize pretreatment,four swarm intelligence optimization algorithms and successive projections algorithm(SPA)were used to extract characteristic variables.The accuracy of the built PLS regression model was satisfied,with the determination coefficient of prediction(Rp)and root mean square error of prediction(RMSEP)were 0.9986 and 0.0628,respectively,indicating that the established model can effectively detect the content of apple juice in the doped samples.(2)Collecting the near-infrared spectroscopy of hemp seed oil doped with soybean oil,peanut oil,sunflower oil and corm oil and applying competitive adaptive reweighted sampling(CARS)to establish the PLS regression model.The CARS-PLS regression model was established to detect the content of the hemp oil seed in different doped types of samples.And the RP2 of the model all exceeded 0.9356,indicating that the built model can effectively analyse the content of hemp seed oil in the doped samples.(3)According to the near-infrared spectroscopy of lotus seed powder samples from Fujian and Hunan province,the partial least squares discriminant analysis(PLS-DA)model was constructed.The accuracy rate of the PLS-DA model was 100%.Clustering algorithm model based on near-infrared spectroscopy of lotus seed powder adulterated with sweet potato powder,wheat flour and corn flour can be used to effectively identify lotus seed powder samples mixed with 5%sweet potato powder,wheat flour and corn flour,respectively.Addictionnally,the near-infrared spectroscopy of lotus seeds from different origin combined with the measured data of moisture content,the CARS-PLS regression model were used to predict the water content of lotus seed powder.The performance of CARS-PLS regression model was good,and Rp2 and RMSEP were 0.9991 and 0.0207.(4)The near-infrared spectroscopy of Tetrastigma Hemsleyanum from five producing areas,including Zhejiang,Fujian,Yunnan,Hunan,and Guizhou province,were collected to construct the PLS-DA classifier model.The identification accuracy was 100%.And the PLS regression model was established based on the near infrared spectroscopy of Tetrastigma Hemsleyanum from different producing areas and the corresponding flavonoids and polysaccharides content.The Rp2 and RMSEP of the CARS-PLS regression model for the flavonoids content were 0.9990 and 0.9287.Besides,the Rp2 and RMSEP of CARS-PLS regression model for polysaccharides content were 0.9998 and 0.0034,demonstrating that near-infrared spectroscopy can effectively identify the origin of Tetrastigma Hemsleyanum and accurately detect the content of flavonoids and polysaccharides.(5)Collecting the near-infrared spectroscopy of Panax pseudoginseng from different grades to construct a CARS-PLS-DA classifier model.The discriminant accuracy of grade B(30 tou),grade C(40 tou)and grade D(60 tou)were all 100%,and the discriminant accuracy rate of grade A(20 tou)was 75%.And the results show that the CARS-PLS-DA classifier discriminant model can be used to fast detect Panax pseudoginseng of different qualities.Through the above research,the built near-infrared spectroscopy analytical model construct foundation for the quality control of daily food,and provided new ideas for the quality identification of related substances through rapid analysis of the substance content.