Research On Underground Target Body Identification Method Based On GPR Sparse Time-frequency Decomposition

As geological conditions become more complex,geological targets change,making exploration more difficult.In order to use ground penetrating radar(GPR)data more rationally and to extract key geological information from radar data,radar data processing methods need to be studied.In this paper,we seek a more efficient method for identifying underground target bodies such as pipes and voids.Starting from the conventional time-frequency analysis method,we focus on the timefrequency analysis method based on the mode decomposition method,and then expand it into a new method for processing radar data by combining the algorithmic properties of energy tracking focus.And its method is effectively applied to the forward model data and the field data,which provides a new basis for the prediction of underground target bodies such as voids and pipes.The main content of this study are as follows,1.This paper introduces a GPR signal processing method based on mode decomposition,and the core algorithm is Multivariate Variational Mode Decomposition(MVMD),which provides a new idea for the decomposition of GPR signals.The principles and underlying theories of Empirical Mode Decomposition(EMD),Ensemble Empirical Mode Decomposition(EEMD),Complete Ensemble Empirical Mode Decomposition(CEEMD),Variational Mode Decomposition(VMD)and MVMD are explained.The above-mentioned mode decomposition methods are applied to synthetic signal and forward model data processing.The comparative analysis shows that the method based on the expansion of VMD to MVMD can meet the demand of decomposing multiple channels at the same time,and avoid the problems of mode mixing and endpoint effects of EMD-like algorithms,and avoid the problem of frequency mismatch,because the decomposition of this method is based on the common frequency components existing among all channels.Therefore,a joint time-frequency analysis method based on the advantages of MVMD is proposed in this paper,which extracts efficient information by adding new energy attributes to achieve the goal of improving the resolution of ground target detection.2.After decomposing the radar data to obtain IMF-slices based on the advantage of MVMD,Teager-Kaiser energy operator(TKEO)is used to compute.Since the TK energy operator is very sensitive to energy tracking and has a better performance on aggregation,it can achieve the purpose of accurate localization and identification subsurface target bodies.Compared with the traditional Hilbert transform-based transient energy slicing,although it can also effectively locate the subsurface target body to extract effective information,the recognition resolution is lower,especially the bottom appears more multiple waves interfere with the identification of the subject target body.The proposed method can effectively suppress multiple waves and highlight the location of the subsurface target body with higher resolution.Finally,we demonstrate the effectiveness of the method using data from a forward model and field data.3.Extending the TK energy operator into a two-dimensional(2D)TK energy operator.Compared with the 1D TK energy operator,which only computes three adjacent samples,the 2D-TKEO takes into account the longitudinal and diagonal feature information,so the 2D-TKEO has stronger stability and reveals the geological information more comprehensively and effectively.Therefore,in this paper,we take advantage of the decomposition of MVMD to obtain the instantaneous spectrum using Hilbert transform and introduce the 2D-TKEO to form a new time-frequency analysis method with higher resolution,which provides a new basis for radar data processing and prediction of subsurface target bodies.We also apply this method to the forward model data and field data processing,and demonstrate that this method has higher resolution than the traditional timefrequency analysis method for locating and characterizing subsurface targets and geological layers.