Abstract: The classification model of adulterated honey was constructed based on two-dimensional（2D） mid-infrared correlation spectroscopy combined with N-way partial least squares discriminant analysis（NPLS-DA）. Firstly, 30 pure honey samples and 30 adulterated honey with sugar samples were prepared respectively. Then, mid-infrared attenuated total reflectance spectra of all samples were obtained in the region of 650-4 000 cm^-1 under room temperature. Spectral features of pure honey and adulterated honey were studied. NPLS-DA model was built using 2D mid-infrared correlation spectra. For comparison, the partial least squares discriminant analysis（PLS-DA）model was built using one-dimensional（1D）mid-infrared spectra. The classification accuracies of two classification models for prediction set were 95% and 90%, respectively. The results show that NPLS-DA model can effectively extract feature information of adulterated honey using 2D mid-infrared correlation spectra and provide higher accurate rates.

Key words: two-dimensional mid-infrared correlation spectroscopy, N-way partial least squares discriminant analysis, adulterated honey, sugar