基于稀疏分布特征的井下微地震信号识别与提取方法

井下微震监测获得的地震记录往往包含大量的噪声,记录信噪比很低.有效地震信号的识别与提取是进行后续地震定位等工作之前需要优先解决的问题.经过研究发现,井下水压裂微地震信号具有稀疏分布的特征,而井下环境噪声则具有更多的Gaussian分布特征.为此,本文提出将图像处理领域适宜于稀疏分布信号降噪处理的稀疏码收缩方法应用于井下微震监测数据处理.为解决需要利用与待处理数据中有效信号成分具有相似分布特征的无噪信号序列估算正交基以及计算效率等问题,将原方法与小波变换理论相结合.即通过优选小波基函数作为正交基进行小波变换将信号分解为不同级的小波系数,利用稀疏码收缩方法中对稀疏编码施加的非线性收缩方式作为阈值准则对小波系数进行改造.通过多方面的数值实验证明了该方法在处理地震子波及井下微地震信号方面准确可靠.含噪记录经过处理后有效地震信号的到时、波形、时频谱特征等均能得到良好的识别和恢复.并且该方法具有很强的抗噪能力,当信噪比低至-20~-30db时,仍然能够发挥作用.在处理大量实际井下微震监测数据的过程中,面对多种复杂情况,本方法展现出了计算效率高、计算结果可靠、应用简单等优势,证明了其本身具有实际应用价值,值得进一步的研究和推广.

Downhole microseismic signal recognition and extraction based on sparse distribution features

On account of working environment, downhole microseismic monitoring records usually contain much noise, i.e. having low SNR (signal to noise ratio). Thus recognizing and extracting effective signals is a priority for subsequent work.#br#In the processing mass downhole microseismic monitoring data, upon a careful study we found that the signals of downhole microseismic events triggered by hydraulic fracturing are characterized by sparse distribution. The background noise has a more significant Gaussian distribution. The relationship between signals and noise is additive blending. Therefore, this study suggests to apply the sparse code shrinkage method, which is present in image processing field, to the processing of downhole microseismic monitoring data. To solve the issues of computational efficiency and needing noise-free training data which have similar statistical properties with the signal component of the data to be processed, we combine the original method with wavelet transformation. That is, we select suitable wavelet bases to take the place of orthogonal bases, and utilize the nonlinear shrinkage process mode of the sparse code shrinkage method as the threshold rule of the wavelet threshold de-noising method.#br#Through many times of numeric simulation tests, it is confirmed that the method presented in this paper is accurate and reliable in processing of seismic wavelet and downhole microseismic signals. The information of arrival time, waveforms, and time-frequency spectra of the effective signal can be well recovered from the noised seismic records. In addition, this method possesses strong anti-noise ability. When the SNR is as low as -20~-30 decibels, the method can still work well. In the process of dealing with the actual data, the method has shown its advantages such as high computation efficiency, accurate calculation, and simplicity of usage. All of these verify the method has a high value of practical application.#br#The downhole microseismic signal recognizing and extracting method based on sparse distribution features has theoretical rationality in processing the hydraulic fracturing microseismic signals, which is a kind of signals with a sparse distribution feature and time-dependent frequency feature. Various numeric simulations and tests on real data processing confirmed this method is applicable. It deserves further research and can be used widely.