Abstract：Using chemometrics, a predictive model for rapid nondestructive detection of lamb tenderloin tenderness was established by analyzing 98 lamb meat samples collected in Beijing by portable near-infrared spectroscopy. The inlfuences of smoothing, derivation and signal correction and other pre-treatments on the model were discussed. The results showed that the optimum pretreatments were average centralization Savitzky-Golay (SG) ifrst-order derivative, SG smoothing and orthogonal signal correction. The model was developed by using partial least square (PLS) regression. The standard errors of calibration and prediction were 0.90 and 2.39, respectively. The coefifcient correlation of calibration set (Rc) and prediction set (Rp) were 0.94 and 0.64, respectively and the number of main factors was 4, which illustrated the model had good predictive accuracy and could be applied in the detection of fresh lamb tenderloin tenderness.
2014, Vol. 28 
