Near-infrared Spectroscopy For Rapid Detection Of Protein And Fat Content In Milk

The content of protein and fat in milk is the key quality index, but most of work on the milk composition measurement are placed emphasis on the traditional analytical methods which are time consuming, expensive, destructive and can not meet the on-line control. Near-infrared spectroscopy (NIR) presents several advantages such as rapidity, precision, no need of sample preparation, and nondestructive aspect. In this paper, the author uses Fourier Transform Near-Infrared Spectroscopy to detect the content of protein and fat in milk powder and liquid milk, for searching after the fast measure technique of milk powder and liquid milk. Main research work and conclusions are as follows:(Ⅰ) The measurement of milk powder'protein and fat content with Fourier transform near-infrared spectroscopy (FT-NIR) was studied.101 milk powder samples of 11 manufacturers and 29 different brands were collected for this study, the contents of protein and fat, which the two main nutrition ingredients of milk powder were selected as detecting indexes. The chemical values, which were determined by Kjeldahl method (GB/T 5413.1-1997) and the Rhodes-Gete Li method (GB/T5009.46-2003) respectively, were taken as the reference data for quantity prediction matrix.The near-infrared spectrum of milk powder samples were scanned by AntarisⅡFourier transform near-infrared spectrometer. Monte-Carlo Sampling (MCS) method was employed to detect the outlier before preprocessing the spectroscopy by appropriate pretreatment methods. Competitive adaptive reweighted sampling (CARS) method was employed to select variables which closely related to the nature of the samples. The partial least squares (PLS) calibration models were established by the optimal conditions to predict the content of protein and fat. The best models showed satisfactory predictions as measured by the R2, RMSECV and RMSEP values:protein,0.9966,0.2320 and 0.1513; fat,0.9959,0.3931 and 0.2780, respectively. Internal and external cross-certification test have shown that near-infrared quantitative analysis has high accuracy and could meet the accuracy needs of prediction of protein and fat content of milk powder.(Ⅱ) The measurement of liquid milk'protein and fat content with Fourier transform near-infrared spectroscopy (FT-NIR) was studied.98 liquid milk samples of four different brands such as Mengniu, Yili, Guangmng and Nestle were collected for this study, the contents of protein and fat, which the two main nutrition ingredients of liquid milk were selected as detecting indexes. The chemical values, which were determined by Kjeldahl method and the Rhodes-Gete Li method respectively, were taken as the reference data for quantity prediction matrix.98 liquid milk samples spectroscopy were collected by four different measure techniques. Monte-Carlo Sampling (MCS) method was employed to detect the outlier before preprocessing the spectroscopy by appropriate pretreatment methods, Competitive adaptive reweighted sampling (CARS) method was employed to select variables which closely related to the nature of the samples, The partial least squares (PLS) calibration models were established by the optimal conditions to predict the content of protein and fat for every measure technique. Correlation coefficient (R2) and root-mean-square error of cross-validation (RMSECV) were used to evaluate the quality of the model. The best measure technique for protein and fat were using diffuse reflectance mode collect the samples heated to 40℃then cooled to room temperature, and collect the samples without heat respectively. The correlation coefficient (R2) of the model were 0.9750,0.9951, root-mean-square error of cross-validation (RMSECV) were 0.1948,0.1363 and root-mean-square error of prediction (RMSEP) were 0.1133,0.1401。The results show that the prediction results by CARS combined with PLS were much better than full spectrum; therefore, CARS method can improve the quality of the model effectively. It is indicate that near-infrared spectroscopy with chemometrics method could detect the content of protein and fat in milk powder and liquid milk fast and non-destructively, and the testing process is simple and easy to control than traditional chemical method, which provide a new approach for detecting the content of protein and fat in milk powder and liquid milk.