Borehole Radar Data Processing Based on Empirical Mode Decomposition

摘要

Borehole radar (BHR) is a highly efficient geophysical exploration tool, however, the application is limited due to relatively small signal to noise ratio (SNR) of the testing data. Empirical Mode Decomposition (EMD) is a data processing method, which takes the particular advantage in analyzing the nonlinear and non-stationary data. This paper introduces the fundamentals of EMD method and the complex signal analysis theory, and focuses on the sufficient conditions and specific decomposition steps for signal processing based on the EMD. Then an actually radar profile, which was obtained by a BHR system in a limestone fracture zone, is used as an example to be analyzed in the following processes. Firstly, the original radar profile is preprocessed to avoid the mode confusion and noise interference to the radar echo. Secondly, the EMD method is used to process a single-channel radar data, and analyze the high-low frequency components of the radar signal. Thirdly, the various intrinsic mode of the preprocessing radar profile is also obtained by the EMD. Finally, we reconstruct the intrinsic mode profile, which contains the effective formation information, calculate the complex signals of the reconstruction radar profile using the Hilbert transform (HT), extract the three characteristic parameters such as instantaneous amplitude, instantaneous phase and instantaneous frequency, and draw the separate instantaneous parameters profiles. The results demonstrate that the EMD method has the strong adaptability in processing the BHR signal under a low SNR and the capability to separate the high-low components of the radar echo effectively. The three-instantaneous information produced by the EMD method and complex signal analysis technique, can highlight the abnormal geological feature, provide the theoretical basis and mutual authentication information for evaluating the original radar profile, and avoid the deviation relying solely on a time-distance profile.