【摘要】 针对太赫兹时域光谱系统由于延时线的重合抖动、采样抖动等产生的噪声，提出使用经验模态分解-R/S分析方法对太赫兹光谱信号进行降噪。采集太赫兹时域光谱系统的时域信号，根据EMD算法将信号分解成本征模态函数(IMF);使用R/S分析法分别计算各个IMF的Hurst指数。根据设定的阈值判断是否各个IMF是否存在均值回复的情况。如果IMF存在均值回复的现象，则使用原始信号与IMF信号作差，所得信号即为降噪后的时域信号。实验结果表明，与小波降噪算法相比，EMD-R/S分析算法能够有效地对太赫兹时域光谱信号降噪，能够有效还原太赫兹光谱信号特征。更多还原

【Abstract】 In the terahertz time-domain spectroscopy system, due to the coincidence jitter of the delay line and the noise generated by sampling jitter, the empirical mode decomposition-R/S analysis method is proposed to denoise the terahertz spectral signal. The time domain signal of the terahertz time-domain spectroscopy system was acquired, and the signal was decomposed into a series of intrinsic mode components(IMF) according to the EMD algorithm. The Hurst index of each IMF was calculated using R/S analysis. It was determined whether the respective IMFs had an average reply according to the set threshold. If the IMF had the phenomenon of mean recovery, the original signal was used to make a difference with the IMF signal, and the resulting signal was the time domain signal after noise reduction. The experimental results show that compared with the wavelet denoising algorithm, the EMD-R/S analysis algorithm can effectively denoise the terahertz time-domain spectral signal and effectively reduce the terahertz spectral signal characteristics.更多还原