Rapid Detection For Quality Of Tibetan Ghee Based On Fourier Transform Near-infrared Spectroscopy

Ghee is a unique dairy product in Tibetan areas,people in plateau has satisfied their need for fat by eating ghee-made products.Fat and protein are the main components of ghee.The fat content accounts for more than 80%with more functional fatty acid content than other dairy products.Ghee has both high food value and medicinal value.However,the nutritional quality in ghee is affected frequently by different storage conditions,and the inappropriate storage conditions cause ghee oxidative rancidity and produce substances that are unfavorable to human health.The traditional method for detecting the quality of ghee is easily influenced by internal and external factors.In addition,the operation is complicated and the cost is relatively high.It is of great significance to establish efficient,rapid and accurate analytical methods.Fourier transform near-infrared（FTNIR）spectroscopy has the advantages of rapidity,convenience,pollution-free and good reproducibility.Therefore,near-infrared technology combined with chemometrics methods were employed for rapid determination of fat contents,protein contents,moisture contents and peroxide value（PV）of ghee to achieve the rapid detection of ghee quality levels.The results are as follows:（1）A total of 54 ghee samples which collected from different regions of Tibet were used to determine the content of fat,protein and moisture,fatty acid composition and its content,and peroxide value using the China national standard method,simultaneously,the statistical analysis was conducted.The results showed that fat in ghee was the main component,content of which was more than 80 g/100g,the protein content ranged from 0.89 g/100g to 1.07g/100g,the moisture content range was form 14.8 g/100g to 20.08 g/100g.The main fatty acids of ghee were oleic acid,palmitoleic acid,stearic acid and tetraoleic acid,the main unsaturated fatty acid was palmitic acid,oleic acid with an average content of 25.98 g/100g.The minimum value of peroxide value was 0.07 mmol/kg and the highest was 2.95mmol/kg.（2）Based on FTNIR,a quantitative analysis model for fat and protein in ghee was established.The results showed that the partial least square method（PLS）quantification model of fats and proteins in ghee was established with the spectral characteristic bands of components(fat:9500-9000 cm^-1,7500-7000 cm^-1,5000-4500 cm^-1,protein:10800-10400cm^-1,7200-6800 cm^-1,6000-5600 cm^-1,and 4400-4000 cm¹)and spectral preprocessing of SNV+second derivative+S-G filter.The correlation coefficients of the model were 0.994（fat）and 0.997（protein）respectively,the root mean square error of the cross-validation was4.09%（fat）and 0.286%（protein）,the number of principal components were 6（fat）and 7（protein）,the model was well-fitted,and the chemical value and prediction of the calibration samples High value correlation.The results of t test showed that there was no significant difference between the predicted and chemical values of the validation samples（p>0.05）,which indicated that FTNIR technology can be applied in detection of content of fat and protein in ghee.（3）The optimal parameters for establishing quantitative model of moisture based on FTNIR spectroscopy are as follow:the spectral ranges were 6665⁷198 cm^-11 and 5066-5599cm^-1.The best preprocessing was SNV+S-G filtering method+first derivative method.The correlation coefficient（r）of the model established was 0.917 and the cross-validation root mean square error（RMSECV）was 2.21%.After the optimization,the spectral variable was reduced from the full spectrum to two band range that 6665-7198 cm^-11 and 5066-5599 cm^-1,which lead to the operation speed of the model was improved.The t test showed that there was no significant difference between the predicted and measured values of FTNIR spectroscopy,indicating that this model is applicable to the quantitative determination of moisture content in ghee.（4）The optimal parameters for establishing a quantitative model of ghee peroxide value based on near infrared spectroscopy are as follows:the raw spectrum is preprocessed by the standard normal variable transform combined with the second derivative and the Savitazky-Golay filtering method.Interval partial least square method（SiPLS）selected four highly correlation spectral bands:12000 to 11467 cm^-1,10933 to 10400 cm^-1,7200 to 6666cm^-1,and 6133 to 5600 cm^-1.The correlation coefficient（r）and the cross validation root mean square error（RMSECV）of optimized model were 0.957 and 0.834%.The model was verified by t test for its performance.The results showed that p=0.324>0.05,indicating that there was a good linear relationship between the model predictions and the measured values.The FTNIR spectroscopy quantitative analysis of ghee peroxide value（PV）is feasible.