基于经验模态分解方法的拉曼光谱信号处理研究

经验模态分解(EMD)方法是一个以信号内在物理频率成分为对象的自适应时频分析方法,而常见的非平稳信号分析方法,比如小波分析,它需要选择小波基,不同小波基的分析结果不同;拉曼光谱信号是典型的非线性和非平稳信号,EMD方法充分地保留了信号本身的非线性和非平稳的特征,适应于拉曼光谱信号的分析。实验在自行研制的拉曼光谱测试平台上获得了原始的拉曼光谱信号,并通过经验模态分解将信号分解成不同频率的10个本征模式分量,信号能量集中在750cm^-1和1500cm^-1左右,最后进行了频率成分分析和去噪处理,并和小波分析方法进行了对比,验证了EMD方法的有效性和实用性,该方法在拉曼光谱信号分析中有较好的应用前景。

Study on signal processing of Raman spectrum based onempirical mode decomposition method

Empirical mode decomposition (EMD) method was an adaptive time-frequency analysis method which took the inherent physical frequency components of signals as the objective. The common non-stationary signal analysis methods such as wavelet analysis needed to select wavelet basis, and the results of various wavelet bases were different. Raman spectrum signal was a typical non-linear and non-stationary signal. EMD method fully retained the characteristics of nonlinearity and nonstationarity characteristics of the signal itself, and it was suitable for the analysis of Raman spectrum signals. The original Raman spectrum signals were obtained on the self-developed Raman spectrum testing platform. The signals were decomposed into 10 intrinsic mode function (IMF) components with different frequencies using EMD method. The energy of signals was concentrated around 750cm^-1 and 1500cm^-1. Finally, the frequency component analysis and denoising treatment were conducted. The effectiveness of EMD method was verified by comparing with the wavelet analysis method. The proposed method had a good application prospect in signal analysis of Raman spectrum.