Time-Frequency Characteristic Analysis of Raman Spectrum Signal

Raman spectrum signal is a kind of scattering signal based on molecular vibration. It is a typical non-stationary signal and considering the scattering frequency shift, the effective information of Raman spectrum is mainly concentrated in the high frequency band. But Raman scattering is very weak, the effective signal is easily disturbed by high frequency noise and low frequency fluorescence background, in order to obtain effective Raman signal, it is necessary to process the signal. Empirical Mode Decomposition (EMD), appears to be a novel signal processing method for nonlinear and non-stationary signals, by decomposing signal into a small number of independent Intrinsic Modes Functions (IMFs) based on scale separation. Ensemble EMD (EEMD) is a substantial improvement of EMD which can better separate the scales naturally by adding white noise signals to the original signals and then treating the ensemble averages as the true intrinsic modes. In this paper, we extend EEMD to Raman spectrum signal analysis. Raman spectrum signals are decomposed and processed by EEMD, and a total of 10 orders IMFs from high frequency to frequency are obtained. According to the energy distribution and amplitude characteristics of the signal, we enhance and reconstruct the characteristics of the effective signal components, and also, we use Partial Least Squares (PLS) method to deduct the baseline of the signal. Through the above processing, the characteristic of vibration mode of the corresponding molecule group is significantly provided, which provides an effective means for further Raman information recognition.