Near-infrared And Mid-infrared Spectroscopic Analysis On Three Nitrogen-rich Illegal Compounds In Soybean Meal

Soybean meal is the main protein feed material in feedstuffs and protein content is the main nutrition indicator of the protein feed, which has great effects on the price of soybean meal. Some enterprises add nonprotein adulterants to seek profit. However, these substances usually cannot be detected by the conventional detection method. The research put forward a new idea about using near- infrared and mid-infrared spectroscopic technology to detect nitrogen-rich adulterants in soybean meal and provides a reference for other more nonprotein nitrogen-rich adulterant analysis in protein feedstuffs based on spectral information. It is of great significant for promoting the development of feed quick analysis technology.We found many nitrogen-rich compounds adulterated in soybean meal after investigation. Three compounds(melamine formaldehyde resin, urea formaldehyde resin and urea polymer) were chosen to study the feasibility of using infrared technology to detect adulterated soybean meal. Near infrared and mid-infrared analysis models were built to detect these substances in soybean meal. The specific research results were as followed. 1. The application of near-infrared diffuse reflectance spectroscopy to detect nitrogen-rich adulterants in soybean mealA total of 196 representative soybean meal samples were collected in this study. Then randomly selected soybean meal samples were mixed with different mass adulterants. The qualitative analysis used partial least square discriminant and support vector machine method and quantitative analysis used partial least square in this study.(1) The soybean meal samples were adulterated with urea polymer from 0.1% to 5.0%, 91 adulterated samples. The identification rate and detection limit of qualitative model was 98.60% and 0.1%, respectively. The determination coefficient, root mean square error and relative percent deviation of quantitative model were 0.9968, 0.0565 and 17.8 in validation set, respectively.(2) The soybean meal samples were adulterated with melamine formaldehyde resin from 0.5% to 5.0%, 84 adulterated samples. The identification rate and detection limit of qualitative model was 98.10% and 0.5%, respectively. The determination coefficient, root mean square error and relative percent deviation of quantitative model were 0.9982, 0.0582 and 23.7 in validation set, respectively. 0.9964.(3) The soybean meal samples were adulterated with urea formaldehyde resin from 0.5% to 5.0%, 94 adulterated samples and 196 authentic samples. The identification rate and detection limit of qualitative model was 98.61% and 0.5%, respectively. The determination coefficient, root mean square error and relative percent deviation of quantitative model were 0.9938, 0.11 and 12.7 in calibration set and 0.9962, 0.091 and 16.5 in validation set. 2. The application of mid-infrared transmission spectroscopy to detect nitrogen-rich adulterants in soybean mealA total of 91 representative soybean meal samples were collected in this study. Then randomly selected soybean meal samples were mixed with different mass adulterants. The qualitative analysis used partial least square discriminant and quantitative analysis used partial least squares method in this study.(1) The soybean meal samples were adulterated with urea polymer from 0.08% to 3.0%, 90 adulterated samples. The identification rate and detection limit of qualitative model was 100% and 0.08%, respectively. The determination coefficient, root mean square error and relative percent deviation of quantitative model were 0.9081, 0.30 and 3.28 in validation set.(2) The soybean meal samples were adulterated with melamine formaldehyde resin from 0.5% to 3.5%, 84 adulterated samples. The identification rate and detection limit of qualitative model was 100% and 0.5%, respectively. The determination coefficient, root mean square error and relative percent deviation of quantitative model were 0.9517, 0.23 and3.48 in validation set.(3) The soybean meal samples were adulterated with urea formaldehyde resin from 0.5% to 3.0%, 77 adulterated samples. The identification rate and detection limit of qualitative model was 98.80% and 0.5%, respectively. The determination coefficient, root mean square error and relative percent deviation of quantitative model were 2.04 in validation set. 3. Comparation of near-infrared and mid-infrared technologyThe concentration of adulterants in soybean meal was correctly predicted by near-infrared technology, whose coefficients of determination were all exceed 0.99 and relative percent deviations were greater than three. So these models could be applied in practice. While mid-infrared technology could predict the concentration of melamine formaldehyde resin and urea polymer; urea formaldehyde resin need further study. Both of two technologies could be applied to discriminate nitrogen-rich additives in soybean meal. But mid-infrared was superior to near-infrared.