摘要: A rapid method based on near-infrared spectroscopy(NIRS) was developed for the measurement of benzoapyrene(BaP) yield in mainstream smoke generated from flue-cured tobaccos. The diffuse reflectance spectroscopy on 167 tobacco samples were collected and pretreated by first-order differential, second-order differential and Savitzky-Golay filter smoothing treatments. The partial least squares (PLS) regression method was adopted to predict the BaP yield from the NIR spectroscopy results. The BaP yield prediction model was established with coefficient of 0.814 and root mean square error of cross validation (RMSECV) of 1.68. Sixteen representative tobacco samples were used as the validation samples for the model. A high degree of correlation between the predicted and the measured values were observed. For BaP yields, the average absolute deviation between the prediction and true value was 1.29 ng/g, and the average relative deviation was 9.88%, with a range of 0.8%-15.8%. T-test results indicated that no significant difference was found between the prediction and test values. This technology does not require complicated smoke collection and analysis by chromatography. Thus, the developed model could be potentially used as a screening tool in the rapid prediction of BaP yield in the flue-cured tobacco smoke.