| Ground penetrating radar(GPR)has been widely used in geophysics.Ground penetrating radar(GPR)transmitting electromagnetic wave signals underground will cause scattering and dispersion problems,which will interfere with target recognition.Moreover,radar echo signal is a kind of non-linear and non-stationary signal.Conventional processing methods can not meet the requirements of detection accuracy.Empirical mode decomposition(EMD),ensemble empirical mode decomposition(EEMD)and complete set empirical mode decomposition(CEEMD)are commonly used to process ground penetrating radar(GPR)measured data,which can decompose radar echo signal into several intrinsic mode functions(IMFs)containing different frequency components of radar signals.Then the low-frequency and high-frequency noises are removed and the remaining IMF components are reconstructed.However,EMD has some shortcomings,such as lack of strict mathematical model,mode aliasing and noise sensitivity.EEMD and CEMD are improved algorithms for EMD,but these problems still exist.In 2014,Variational Mode Decomposition(VMD)proposed by Dragomiretskiy.K.et al.is a completely non-recursive signal decomposition method based on frequency domain,which overcomes many shortcomings of EMD to a certain extent.By iteratively searching the optimal solution of the variational model,the frequency center and bandwidth of each decomposition component can be determined,and the frequency domain division and effective separation of each component can be realized adaptively.Compared with the EMD,EEMD and CEMD,VMD converts signal decomposition into non-recursive and variational mode decomposition modes,showing better noise robustness.Because the VMD method has just been put forward and has not been popularized,its application in practical engineering has been rarely reported.In this paper,it is applied to the data processing of GPR to analyze the useful target signals.The signal processing results of VMD algorithm are affected by both penalty parameters and the number of components.In order to separate the signal components containing abundant characteristic information from the ground penetrating radar signals with low signal-to-noise ratio and achieve the best processing effect,this paper uses Particle Swarm Optimization to search for the best combination of influence parameters of VMD algorithm,and proposes a data processing method based on parameter optimization VMD.Independent component analysis(ICA)is a new signal processing method based on high-order statistical information of samples.It has been widely used in communication signal processing,audio signal processing and other fields.In order to solve the problem of denoising and reconstruction of IMF components,a variational mode decomposition denoising method based on parameter optimization and kurtosis comparison is proposed.Firstly,two adjacent records of ground penetrating radar signal are taken,and the effective signal of peak value GPR and the random noise signal of low peak value are obtained by independent component analysis method.The phase of peak value signal is judged and the automatic inverse phase correction is carried out.Then,the optimal parameter combination of variational mode decomposition algorithm is searched by particle swarm optimization algorithm,and the peak value signal is changed.Several IMF components are obtained by mode decomposition,and the peak value of each IMF component is calculated.Finally,the peak value of noise obtained by independent component analysis is taken as the threshold value,and the IMF component whose peak value is higher than the threshold value is taken as the signal component.The reconstructed signal is accumulated and the denoising process is completed.Through the analysis of the forward simulation data and the measured data of ground penetrating radar,the results show that the proposed method can effectively remove the noise and interference in ground penetrating radar signal,and retain the effective signal of the target body. |
