KANG Jing, YAO Haibo, LIANG Ting, TANG Defu, NIAN Fang
In order to rapidly determine the nutrient content of a large number of fresh lamb samples, this study used near-infrared reflectance spectroscopy (NIRS) to construct a quantitative analysis model of six nutrients in fresh lamb, Totally 203 fresh mutton samples were collected in Minqin county, Wuwei city, and moisture content (moisture, MT), crude fat (ether extract, EE), crude protein (crude protein, CP), glucose (glucose, Glu), crude ash (crude ash, Ash) and total phosphorus (total phosphorus, P) content of samples were measured, WINISI III and Foss Calibrator calibration software were used to establish NIRS models of 6 kinds of mutton nutrients and the results were compared, WINISI III software calibration results showed that the coefficient of determination for validation (RSQ) and the ratio of performance to deviation for validation (RPD) of the MT, EE, and CP prediction models were 0.83 and 2.47, 0.90 and 3.60, 0.81 and 2.79 respectively;the RSQ and RPD of the Glu and Ash prediction models were 0.54 and 3.05, 0.54 and 1.91, respectively;the RSQ and RPD of the P predictive model were 0.45 and1.80, respectively, The calibration results of Foss Calibrator software showed that the root means a square error of cross-verification [RMSEP(cross)] and coefficient of determination (R2)of MT, EE, and CP were 0.631 and 0.84, 0.326 and 0.87, 0.468 and 0.83, respectively;RMSEP (cross) and R2 of Glu and Ash were 0.127 and 0.53, 0.179 and 0.51, respectively; RMSEP (cross) and R2 of P were 0.086 and 0.33, respectively, The conclusions obtained by the two-calibration software were similar, indicating that the prediction models of MT, EE, and CP can be accurately predicted in actual production, The prediction models of Glu and Ash can be used to roughly analyze and screen a large number of samples, and need to be further optimized. The prediction model of P had a poor correlation and cannot be applied in actual production.
