Abstract: Both qualitative and quantitative detection models were established for thawed beef patties (with different proportions of fat) adulterated with pork using near infrared spectroscopy combined with chemometrics. The results showed that for the qualitative determination of samples with different proportions of adulteration, partial least squares discriminant analysis was better than principal component analysis (PCA) combined with support vector machine method, and the discriminant analysis gave an accuracy of 100% for both calibration and validation sets. Partial least squares regression method was applied to quantitative detection of thawed beef patties with different proportions of adulteration. The correlation coefficient of calibration (Rc) and prediction (Rp) and the root mean square error of prediction (RMSECP) of the model were 0.968 9, 0.861 1 and 7.221%, respectively. Therefore, near infrared spectroscopy (NIS) can be applied to both qualitative and quantitative detection of thawed beef patties with different proportions of fat adulterated with different amounts of pork.

Key words: near infrared spectroscopy, beef patties, pork, adulteration, qualitative and quantitative detection