Model Establishment For Predicting The Chemical Composition Of Soybean Meal And DDGS By Near Infrared Spectroscopy

One hundred soybean meal samples and DDGS samples were collected. Based on the moisture(H2O),crude protein(CP),ether extract(EE),crude ash(Ash) and crude fiber (CF)of these samples analysis, the samples were assayed by InfraXact near infrared spectroscopy to establish the calibration models of H2O, CP, EE, Ash and CF in soybean meal and DDGS.The results as follows:1. The NIRS calibration models of soybean meal were developed using the Modified Partial Least-Squares(MPLS)regression technique, except that of the EE.Initial spectrum was pretreated, MSC were the best method in H2O,CP and EE, while SNV AND DETREND were the best method in Ash and CF.2. The calibration model of H2O,CP,EE,Ash and CF in Soybean meal under 40 mesh,the RSQ were 0.9811,0.9475,0.8716,0.7300 and 0.9601 , SEC were 0.0626,0.2880,0.1354,0.1199 and 0.2575,RSD were 0.4861%, 0.6536%, 12.321%, 2.0489% and 5.5811%,RPD were 7.2668,4.3660,2.791,1.9250 and 5.0050, respectively.The crossing and the external inspection RSQ were above 0.90,except that of the EE and the Ash . the EE and the Ash need further improve the model accuracy.3. The NIRS calibration models of DDGS were developed using the Modified Partial Least-Squares(MPLS)regression technique.Initial spectrum was pretreated,SNV AND DETREND were the best method,except that of the EE.4. The calibration model of H2O,CP,EE,Ash and CF in DDGS under 40 mesh, the RSQ were 0.9873,0.9717,0.9959,0.9636 and 0.9684,SEC were 0.1816, 0.4775, 0.2143,0.1607 and 0.4043 , RSD were 2.0391%,1.6172%,3.4677%,3.7906% and 4.6690%,RPD were 8.8645,5.9468,15.5623,5.2433 and 5.6270,respectively.The crossing and the external inspection RSQ were above 0.90.Conclusion: The results show the models could be well applied on chemical contents prediction in soybean meal except that of the EE and Ash. The calibration models to predict the H2O,CP,EE, Ash and CF in DDGS could meet the needs of quality evaluation.