Rapid identification of milk samples by high and low frequency unfolded partial least squares discriminant analysis combined with near infrared spectroscopy
writer：Xihui Bian*, Caixia Zhang, Peng Liu, Junfu Wei, Xiaoyao Tan, Ligang Lin, Na Chang, Yugao Guo
keywords：Empirical mode decomposition, Partial least squares discriminant analysis, Near-infrared spectroscopy, Milk identification,Pattern recognition
source：期刊
specific source：Chemometrics and Intelligent Laboratory Systems, 2017, 170, 96-101
Issue time：2017年
A high and low frequency unfolded partial least squares discriminant analysis (HLFUPLS-DA) for building a pattern recognition model of near-infrared (NIR) spectra is proposed to identify milk samples. In the approach, the spectra are decomposed into different frequency components by empirical mode decomposition (EMD) at first. Then the former high frequency components are summed as a high frequency matrix and vice versa. Thirdly, the high and low frequency matrices are extended to an extended matrix in the variable dimension. Finally, PLS-DA model is built between the extended matrix and the target vectors. Coupled with NIR spectroscopy, HLUPLS-DA is applied to identify milk samples of different qualities. Comparing with PLS-DA and other signal processing techniques combined with PLS-DA, the proposed method is proved to be a promising tool for spectral qualitative analysis of complex samples.