Research On Noise Suppression Of Microseismic Data Based On Higher-order Cumulant And Shearlet Transform

In recent years, the development of low permeability oil and gas reservoirs is the main object of future oil and gas exploration in China and the inevitable trend of China’s oil and gas industry development in the future. Low permeability reservoir exploitation in the process of hydraulic fracturing will induce a series of microseismic events and the microseismic signal monitoring has been widely used in the evaluation of artificial fracturing effect. The actual microseismic record is complex and the effective signal is hard to be recognized because of noise. Noise suppression of microseismic data is the premise of subsequent processing. With the continuous development of the mathematical theory of signal processing, more and more methods are applied to the microseismic noise suppression. In this thesis, firstly the actual microseismic noise is analysed, then high-order cumulant theory and Shearlet transform theory are used to suppress microseismic noise based on research background at home and abroad.In this thesis, firstly the characteristics of the effective signal and the actual microseismic noise are summarized. In view of the actual noise data in Shanxi, the probability statistical analysis method is used to prove that most of the noise satisfies the Gauss property, which provides a theoretical basis for the subsequent theoretical method.Then the basic principles of highr-order statistics and Shearlet transform are introduced systematically. The advantages of them are analyzed about the noise suppression of microseismic data in detail.In view of the deficiency of the traditional cross correlation method, the higher-order cumulant theory is introduced. A method is proposed to obtain the time delay estimation based on the one dimension slice of the three-order cumulants. Firstly, the three-order self accumulation and mutual accumulation of the signals are calculated, then one dimensional slice is gained in order to reduce the computation and satisfy the field real-time requirements, finally, the time delay value is obtained by the criterion function. This method is not sensitive to the unknown Gauss noise, especially for the correlated Gauss noise. The simulation results show that the time delay detection probability and the root mean square error are better than the cross correlation method under the influence of the random Gauss noise and correlated Gauss noise. In the implementation of real data we select the data from the well as a reference, then calculate the higher-order cumulant with the ground data based on surface and borehole observation. The validity and practicability of the method is verified.In view of the deficiency of Wavelet and Curvelet transform, the Shearlet transform theory is introduced, and the denoising method of the microseismic data based on the non-subsampled Shearlet transform is proposed. Compared to other multi-scale transform methods, it has optimal sparse representation, better direction sensitivity, stronger ability to remove noise and better signal fidelity. At the same time, it has the translation invariance and eliminates the pseudo gibbs phenomenon. Simulation and real data are analyzed by using the non-subsampled Shearlet transform, Wavelet and Curvelet transform threshold denoising method. Results show that the non-subsampled Shearlet transform has better denoising ability.Finally, the Shearlet transform and higher-order cumulant theory are combined. Firstly the Shearlet transform is used to preprocess the microseismic data, and then the accurate time delay values are obtained by the higher-order cumulant operation after further suppress noise. The effectiveness is verified by simulation and real data processing.