Microseismic Signal Denoising Method Based On EM-KF Algorithm

Hydraulic fracturing is a reservoir modification technology for increasing production of oil and gas wells and increasing injection of water injection wells.Microseismic events generated during hydraulic fracturing can be collected using microseismic monitoring techniques.During microseismic monitoring,the path of seismic wave travel is long and the signal attenuation is large.In addition,the noise types of microseismic data are complex,and the overall signal-to-noise ratio of microseismic data is low,which brings great difficulties to subsequent first break picking and inversion positioning.Therefore,it is necessary to conduct noise removal processing for the noise interference existing in microseismic monitoring data.The denoising processing of microseismic signals can eliminate noise interference in microseismic signals,thereby better identifying effective information in microseismic signals.This can improve the acquisition quality and accuracy of microseismic monitoring data,enabling us to more accurately analyze underground structures and rock stress states.At the same time,denoising processing can also make the data clearer and easier to read,facilitating subsequent processing and analysis.Therefore,denoising of microseismic signals is a necessary process.Model based microseismic denoising is a common denoising technique,whose goal is to establish an appropriate mathematical model to describe the observed data and use the model to remove noise.Model based denoising methods have the advantages of wide application range,stable denoising effect,and strong reliability.Therefore,model based methods are selected to denoise microseismic signals.This article will conduct research from the following three aspects.Firstly,this paper analyzes and studies the characteristics of microseismic signals and noise.According to the characteristics of microseismic signals,forward synthesis of simulated microseismic signals is performed.According to the characteristics of noise,add appropriate noise to simulate real noise.According to the propagation characteristics of microseisms,establish a state space equation that conforms to the propagation laws of microseismic signals,so as to facilitate subsequent denoising of microseismic signals.Secondly,we studied the denoising method of microseismic signals based on EM-KF(Expectation Maximization Kalman Filter).The combination of EM algorithm and Kalman filter is selected for denoising.Kalman filter has the characteristics of small computational complexity,high real-time performance,and good denoising effect.Kalman filter can be used for denoising microseismic signals.In combination with EM algorithm,Kalman filter can be used for parameter estimation to improve the denoising performance of Kalman filter.Thirdly,a denoising method for microseismic signals based on EM-KF convolutional neural networks was studied.Based on model based denoising methods,the quality of model selection has a significant impact on the denoising results.If the noise has a significant impact on the signal,the denoising effect will be affected.Combining the EM-KF algorithm with convolutional networks to achieve a better denoising effect.This article proposes two denoising algorithm models for microseismic signals.The use of EM-KF algorithm and EM-KF based convolutional neural network methods effectively improves the signal-to-noise ratio of microseismic signals,providing high signal-to-noise ratio microseismic signals for subsequent work.