Fluorescence spectroscopy recognition of mineral oil based on lifting wavelet- BP-fuzzy neural network

There is a lot of noise in the original spectral signal collected from the detector. In this paper, the lifting wavelet was used to remove the noise. The singular value eigenvector obtained by the Excitation-Emission Matrix (EEM) factorization from the fluorescence spectroscopy was used to recognize the information of the mineral oil. By the BP-Fuzzy Neural Network the singular value eigenvector was trained and the recognition of the many kinds of mineral oils was realized. The experiment results show that it is effective to remove the noise in the spectral signal and hold the useful local signal. It can effectively recognize the fine distinction between the different spectrums and realize the identification of the oils.