EMD-based denoising of multi-transient electromagnetic(MTEM)signal and its time-frequency(TF)processing in 2-D and 3-D using generalized Stockwell transform(GST)and dynamic autoregression(DAR)model
Multi-transient electromagnetic method(MTEM)is a new effective technique usingpseudo-random binary sequence(PRBS)as a source waveform to study geo-electric structure.The depth of investigation is determined by the signal-to-noise ratio(S/N).MTEM field datais contaminated by various noise sources which may result in erroneous interpretation.Toovercome this problem,we propose a new approach for MTEM data processing based onEmpirical Mode Decomposition(EMD)denoising algorithm coupled with time-frequency(TF)distribution to increase penetration depth by improving SNR.We contaminated thePRBS source signal at three SNR levels(-1,3,10 dB).It was decomposed by the EMD toobtain a series of intrinsic mode functions(IMFs).The interval hard thresholding thenremoved the noise-only IMFs and reconstructs the estimated signal with the processed IMFsto get the de-noised TEM signal.The spectral analysis are carried out by GeneralizedStockwell transform(GST)and Dynamic AutoRegression(DAR)model as a 2-D & 3-D TFanalysis tool.The proposed approach allowed to suppress mixed electromagnetic noise andenhanced denoising performance is evaluated using quantitative indicators such as low signaldistortion ratio(SDR)and high noise suppression ratio(NSR)and energy concentrationmeasure(ECM)for reliable TEM data interpretation.
Time-Frequencyrepresentation Non-stationary Time series Multi-Transient Electromagnetic method(MTEM) Empirical Mode Decomposition(EMD) Generalized Stockwell transform(GST) Dynamic AutoRegression(DAR)
Muhammad Younis Khan Guoqiang Xue Li Hai Gang Yu Weiying Chen
Key Laboratory of Mineral Resources,Institute of Geology and Geophysics,Chinese Academy ofSciences,B School of Electrical Engineering,University of Jinan 250022,China
国内会议
武汉
英文
398-402
2017-11-10（万方平台首次上网日期，不代表论文的发表时间）