The Research Of FWI Based On EMD Decomposition

Full waveform inversion(FWI) can construct a high-resolution velocity model,which reconstructs the subsurface velocity structure based on the kinematics and dynamics information of the prestack seismic wavefield,and can reveal the structural and lithological details of the complex geological background.According to actual needs,full waveform inversion can be realized in time,frequency and Laplacian domain.From the point of view of calculation methods,the implementation methods of full waveform inversion mainly include gradient guidance method and Newton method(Gaussian Newton method).However,there are still many problems in the actual exploration data processing: due to the complexity of the underground wave field,the nonlinearity of the inversion problem will be enhanced,and it will fall into the local extremum in the process of optimization;the conventional full waveform inversion cannot To solve the problem of low frequency data loss,when the seismic data lacks low frequency information,how to recover the long wavelength component of the underground medium is the most important issue.In order to solve this problem,in this paper,we first derive the relevant formula from the beginning of the forward performance,and give a comparison chart of the PML boundary and the PML-free boundary wave field cannon record.Then,by analyzing the FWI resolution,we give The FWI method's inversion limit of the subsurface medium scale and its relationship with the observation system,according to the conclusion of the resolution analysis,it is proposed to use the Empirical Mode Decomposition(EMD)method to decompose the observation data.The observed data is decomposed into a finite number of Intrinsic Mode Functions(IMFs).The IMF components are separated into local characteristic signals of different time scales of the original signal,so that the waveform corresponding to the underground large-scale structure is obtained,and then Used for waveform inversion.The empirical mode decomposition method is a linear and steady-state spectrum analysis based on Fourier transform.The method performs signal decomposition based on the time scale characteristics of the data itself,without pre-setting any basis functions.Therefore,it has obvious advantages in dealing with non-stationary and nonlinear data.It is suitable for analyzing nonlinear and non-stationary signal sequences and has a high signal-to-noise ratio.Therefore,the empirical mode decomposition method has great advantages in seismic record analysis compared to other methods.At the end of the paper,the salt hill model test shows the effectiveness of the above method.