提出了基于去趋势波动分析（DFA, Detrended Fluctuation Analysis）和小波变换（WT, Wavelet Transform）改进的经验模态分解算法(EMD，Empirical Mode Decomposition)自适应去噪方法。含噪信号经EMD自适应分解成若干本征模态函数(IMF, Intrinsic Mode Function)分量，利用DFA计算各IMF的尺度指数，自适应选出具有长程相关性的低频有用分量，并用WT处理高频分量，最后利用选取的低频有用分量和小波处理后的高频分量重构获得去噪信号。理论分析和实验表明，EMD-DFA和小波去噪两种方法去噪后的输出信噪比分别为12.3458 dB和13.7369 dB，相比单一的降噪算法，改进算法EMD-DFA-WT的输出信噪比为14.4513dB，信噪比得到了明显提高，有效消除了天然气管道泄漏信号中的噪声，突出信号特征，达到了降噪的目的。

An improved Empirical Mode Decomposition（EMD）adaptive denoising algorithm based on Detrended Fluctuation Analysis (DFA) and wavelet transform (WT) is proposed. The noisy signal is adaptively decomposed into several Intrinsic Mode function (IMF) components by EMD, the scale index of each IMF is calculated by DFA, and the low-frequency useful components with long-range correlation are adaptively selected, the high frequency components are processed by WT. Finally, the selected low-frequency useful components and the high-frequency components after wavelet processing are used to reconstruct the denoised signal. Theoretical analysis and experiments show that the output signal-to-noise ratio of EMD-DFA and wavelet denoising methods after denoising are 12.3458dB and 13.7369dB, respectively.Compared with the single denoising algorithm, the output signal-to-noise ratio of the improved algorithm EMD-DFA-WT is 14.4513dB, and the signal-to-noise ratio is significantly improved, which effectively eliminates the noise in the leakage signal of the natural gas pipeline, highlights the signal characteristics, and achieves the purpose of noise reduction.

国家自然科学基金(61873058)；中国石油科技创新基金(2018D-5007-0302)。