微地震数据去噪方法研究

发布时间：2018-08-24 09:11

【摘要】：在最近的几年里,石油的消耗总量逐年增加,低渗透储层油气藏在整个油田的石油产量中所占的比重也在逐渐加大。水力压裂可以改造地层结构,在原本封闭的储层和井筒之间压裂出一条新的裂缝作为流体的新通道。此方法可以大大提高低渗透储层油气藏的出油量,在当前的油气田勘探开发中充当着一个不可或缺的重要角色。在水力压裂施工的同时,需要使用微地震监测技术观测压裂过程中地层裂缝的空间图像,并对水力压裂的压裂效果做出符合实际情况的评判。影响评价结果的重要因素是：通过去噪处理后得到的微震数据信噪比的高低。由于微地震事件持续时间较短、声波频率较高、释放能量较小的特点,在实际生产过程中采集到的微地震数据中掺杂着许多的干扰噪声信号,更有甚者干扰噪声信号完全淹没了有效信号。因此,对采集到的微地震信号在自动拾取有效信号之前做去噪处理是非常有必要的。本文通过对微地震信号中干扰噪声信号的类型进行分析,同时结合最近几年国内外的在微震信号去除噪声方面的技术经验,对几种地震数据去噪方法进行试验与分析,并把改进后的去噪技术运用到微地震数据处理的实例上。本文从当前油田的实际生产现状出发引出了微地震监测在油田开采中后期的重要性,简单介绍了国内外微地震监测方法从提出理论到应用到实际生产的过程、利用检波器采集微地震监测数据的方法(一般为井中监测)、微地震监测这一技术由哪些仪器所组成。从微地震数据干扰信号的类型和特点入手,主要研究了预测反褶积滤波、相关滤波、变模态分解、曲波变换等微震信号去噪方法。最后,使用MATLAB编写了预测反褶积滤波、带通滤波、相关滤波、变模态分解、曲波变换等程序,把前面提到的几种滤波或重构去噪方法在MATLAB的语言平台上对现场采集的微地震数据进行了分析和处理。得到的处理结果表明：(1)在实际生产过程中,裂缝位置的实时精确成像和有效信号的自动拾取与微震资料的信噪比密切相关。(2)微地震信号震源位置的确定是去噪处理的一个关键问题,在微地震信号震源位置未知的情况下,很多完善的地震资料处理方法应用受到极大的限制,不能发挥其真正作用。(3)微震资料中包含有：有效波和干扰波,细分为直达波、折射波、透射波、多次波、转换波、面波、管波和导波；纯纵波和纯横波为线性极化波,面波、导波为非线性极化波。(4)主成分分析对压制微震信号中的随机干扰噪声有一定的效果,压制噪声的效果比带通滤波要好一些。(5)频率滤波器是基于有效信号与噪声在频率域具有可分性这一特征,设定压制某特定频段的信号从而起到压制噪声的效果。因此,频域滤波器的使用条件是,只有当有效信号和噪声在频率域内存在较大差异的时侯,才能起到较好的压制噪声的效果。(6)预测反褶积滤波可以压缩子波并有效去除微震资料中的多次波。预测反褶积滤波在处理包含多次波噪声的微震数据时,不使用速度等辅助信息并且运行速度较快。(7)相关滤波方法是在随机干扰背景上凸出微震信号同相轴比较有效的方法。根据实测微震资料选取合理的互相关参数,可以有效去除随机干扰噪声,提高同相轴分辨率。(8)曲波变换拥有多尺度各向异性的特点,解决了小波变换在处理微震资料边沿的方向特质等方面的内在不足,能够在压制随机干扰信号的同时保留微震信号的细节,实现能更去除噪声的目的。(9)由于微震信号具有其特殊性和复杂性,单一的微震信号去噪方法难见其效,表明单一的方法都有其局限性。多种方法综合使用才是今后微震信号去噪技术探索的方向。
[Abstract]:In recent years, the total consumption of oil has increased year by year, and the proportion of low permeability reservoirs in the oil production of the whole oilfield has gradually increased. Oil production of high and low permeability reservoirs plays an indispensable and important role in the current exploration and development of oil and gas fields. While hydraulic fracturing is being carried out, it is necessary to use microseismic monitoring technology to observe the spatial image of formation fractures in the process of fracturing, and to make a judgment of the fracturing effect in accordance with the actual situation. The important factors influencing the evaluation results are the signal-to-noise ratio of the microseismic data obtained by denoising. Because of the characteristics of short duration of microseismic events, high frequency of sound waves and low energy release, many interference noise signals are mixed with the microseismic data collected in the actual production process, and even more interference noise. The acoustic signal completely submerges the effective signal. Therefore, it is necessary to denoise the collected microseismic signal before picking up the effective signal automatically. In this paper, the types of interference noise signals in the microseismic signal are analyzed, and the technology of removing noise from the microseismic signal at home and abroad in recent years is combined. In this paper, the importance of microseismic monitoring in the middle and late period of oilfield production is introduced from the actual production situation of the current oilfield, and the microseismic monitoring methods at home and abroad are briefly introduced. In this paper, the method of acquiring microseismic monitoring data by geophone (usually in-well monitoring) is discussed, which instruments are used in microseismic monitoring. Starting with the types and characteristics of the interference signals of microseismic data, the main research contents are predictive deconvolution filtering, correlation filtering, variable mode decomposition and curved wave transformation. Finally, the program of predictive deconvolution filtering, band-pass filtering, correlation filtering, variable mode decomposition, curved wave transformation and so on are compiled by MATLAB to analyze and process the microseismic data collected on the spot on the language platform of MATLAB. The results show that: (1) In the actual production process, the real-time accurate imaging of fracture location and the automatic pick-up of effective signals are closely related to the signal-to-noise ratio of microseismic data. (2) The determination of the source location of microseismic signals is a key problem in the de-noising process. The application of this method is greatly limited and can not play its real role. (3) Microseismic data include: effective wave and interference wave, which are subdivided into direct wave, refraction wave, transmission wave, multiple wave, converted wave, surface wave, tube wave and guided wave; pure longitudinal wave and pure shear wave are linear polarized wave, surface wave and guided wave are nonlinear polarized wave. (4) Principal component analysis is used to counter-pressure. (5) Frequency filter is based on the separability of effective signals and noises in the frequency domain. It is set to suppress the signal of a certain frequency band so as to suppress noise. (6) Predictive deconvolution filtering can compress wavelets and effectively remove multiple waves from microseismic data. Predictive deconvolution filtering does not use velocity and other aids in processing microseismic data containing multiple noise. (7) Correlation filtering method is an effective method to convex microseismic signal in-phase axis on random interference background. Selecting reasonable cross-correlation parameters according to measured microseismic data can effectively remove random interference noise and improve the resolution of in-phase axis. (8) Curvilinear transform has the characteristics of multi-scale anisotropy. (9) Because of the particularity and complexity of microseismic signal, it is difficult to denoise a single microseismic signal, which indicates that the denoising method is ineffective. A single method has its limitations. The comprehensive use of multiple methods is the direction of future exploration of microseismic signal denoising technology.
【学位授予单位】：长江大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.44

【参考文献】