Geographical origin traceability and authenticity detection of Chinese red wines based on excitation-emission matrix fluorescence spectroscopy and chemometric methods

Excitation-emission matrix fluorescence (EEMF) spectroscopy combined with chemometrics was developed for the geographical origin traceability and authenticity detection of Chinese red wines. Parallel factor analysis coupled with partial least squares-discriminant analysis (PARAFAC-PLS-DA), N-way partial least squaresdiscriminant analysis (N-PLS-DA) and unfolded partial least squares-discriminant analysis (U-PLS-DA) were used to identify the geographical origins and adulteration of red wines. Accuracy, sensitivity, specificity and precision were used to evaluate the performances of the discriminant models. The accuracy of PARAFAC-PLS-DA for training samples and prediction samples from different origins was 0.88 and 0.83, respectively. N-PLS-DA and U-PLS-DA, which used full EEMF data as model inputs, had the same excellent performance and both provided an accuracy of 1. With regard to the identification of adulterated samples, the accuracy of PARAFAC-PLS-DA for training samples and prediction samples was 0.71 and 0.83, respectively. By comparison, in both N-PLS-DA and U-PLS-DA models, only one prediction sample with the lowest adulteration level was misclassified. N-way partial least squares regression (N-PLSR) was successfully used to quantify adulteration levels for the first time, with R & ccaron; > 0.98, both RMSEC and RMSEP < 6%. This study showed that EEMF combined with chemometrics is an effective method for the geographical origin traceability and authenticity detection of Chinese red wines.