Research On Downhole Microseismic Noise Suppression Method Based On Multi-scale Transformation

As the growing demand for oil and gas in the world,low-permeability reservoirs are becoming an important way to increase oil production.Hydraulic fracturing is an effective means of evaluating low-permeability oil fields.Microseismic monitoring is used as a low-permeability oil and gas field fracturing.Monitoring methods have received much attention.Microseismic monitoring is usually divided into two types: surface and downhole monitoring.Among them,the downhole microseismic records have the characteristics of high frequency,weak energy and low signal-to-noise ratio(SNR).Improving the SNR of monitoring records has become an indispensable step for subsequent signal processing.Therefore,suppressing noise in the recording is an urgent task that needs to be solved.The main research content of this paper is the study of random noise suppression methods downhole microseismic records.A method of achieving noise and signal separation by comparing the original signal with a threshold is called a threshold denoising method.The threshold method is a relatively simple and effective denoising method,but the traditional threshold method has its inherent defects.In this paper,the continuity and error of various threshold functions are studied.It is found that the improved weighted average threshold function has the characteristics of changing with the change of weight,which can reduce the error while maintaining the continuity of the denoised signal.In this paper,the improved weighted average threshold function is combined with the Empirical Wavelet Transform(EWT)and the Three-Dimensional Shearlet Transform(3DST).At the same time,based on the characteristics of downhole microseismic data and the improved weighted average threshold function,the threshold schemes of EWT and 3DST are given respectively.EWT is obtained by mathematical combination of wavelet transforms of different dimensions and Empirical Mode Decomposition(EMD),which improves the shortcomings of WT transform adaptive weak and EMD but less theoretical basis.This paper first gives a threshold denoising scheme for one-dimensional EWT.By analyzing multiple spectrum segmentation methods in EWT,and comparing the time-frequency characteristics of the decomposed signals,an adaptive algorithm more suitable for microseismic signals is selected.The adaptive algorithm can adaptively separate the effective signals and noise of the downhole microseismic data into different modes.Then,the modes are divided into two categories by analyzing the frequency and energy information of each mode.A threshold function with a weight of zero is applied to a mode containing more significant signals,and a threshold functionwith a weight greater than zero is applied to a mode containing less significant signals.By analyzing various threshold estimation methods,a hierarchical unified threshold method which is more suitable for multi-modes characteristics after EWT decomposition is selected.Simulation experiments verify the effectiveness of the one-dimensional EWT threshold denoising algorithm.Second,based on the threshold denoising of one-dimensional EWT,this paper considers the correlation between the microseismic signals,and applies the two-dimensional EWT to the denoising of seismic signals.The simulation results show that the two-dimensional EWT threshold denoising algorithm can better suppress noise.By calculating the number of cross-correlation between three-component(3C)data,it is proved that the 3C data has a certain spatial correlation in three-dimensional space.The EWT transform denoising method does not take into account the correlation of 3C data.Therefore,the3 DST threshold denoising scheme is further given in this paper.Firstly,a mechanism for constructing a three-dimensional matrix using the correlation between 3C data is established.The 3D matrix transform is used to transform the 3D matrix into the shearlet domain to obtain different scale coefficients.Different scales are grouped by analysis of energy and high-order cumulants of each scale factor.Set threshold parameters that are more suitable for 3C data,and denoise using an improved threshold function.The experimental results show that the algorithm significantly improves the SNR of the microseismic data and effectively preserves the useful signal.