基于Fast-AIC算法的微地震事件初至拾取及自动识别技术研究

发布时间：2018-08-15 16:40

【摘要】：微地震监测技术是目前评价油田压裂效果最为可靠的方法之一。通过对微地震事件的准确定位,可以判断裂缝走向、维度以及分析反演震源机制等,为油田的后续生产、开发提供依据。而微地震事件自动识别及初至拾取是该技术的一项关键处理步骤,地震波初至到时的拾取速度和拾取精度直接影响微地震定位效率和定位结果的可靠性。微地震监测检波器布阵方式分为井中监测方式和地面监测方式,地面监测方式不需要监测井,施工简单、成本低,将是未来各大油气田主要应用的微地震监测方式。而微地震震源位置位于地表以下几百米甚至几千米,微地震记录普遍信噪比较低,导致目前已有的微地震事件自动识别及初至拾取方法在处理实际微地震数据时都具有一定的局限性。因此,本课题针对微地震地面监测方式,围绕微地震监测数据的特点等方面,对微地震事件自动识别和初至拾取技术进行了研究。水力压裂过程中诱发的微地震事件数量以及诱发时刻皆不确定,为准确拾取所有微地震事件对应的微地震信号的初至到时,需先确定微地震事件个数和大致诱发时刻,再根据微地震事件诱发时刻在附近选取局部微地震数据进行精确初至拾取。对于微地震事件自动识别方法,目前常规方法都是基于单台地震仪记录的信号进行处理,识别微地震事件时常出现漏检、误检的情况,无法保证每台地震仪记录中识别得到的微地震事件个数一致。Akaike Information Criteria(AIC)方法实现简单、便于计算,尤其适合于含有微地震事件的局部数据拾取,但针对长时间记录的海量微地震数据处理,其计算效率有待提高。针对以上研究中存在的缺陷,结合射孔信号提出开展基于Fast-AIC算法的微地震信号初至拾取方法研究。首先,为了避免基于单道微地震数据的常规微地震事件识别方法中漏检、误检的缺陷,利用监测区域地层速度先验信息,采用道间能量叠加技术提出基于射孔信号的微地震事件自动识别方法。同时,为了克服基于射孔信号的微地震识别方法对前期射孔数据的依赖性,提出当没有实际射孔数据时,根据声波测井曲线建立层状速度模型,采用射线追踪理论正演模拟射孔信号代替实际射孔信号进行计算。然后,根据水力压裂监测系统检波器布置参数及地层速度等参数在微地震事件触发时刻附近选取局部有效数据,并采用Curvelet变换对局部数据进行滤波处理,为精确拾取微地震信号提供高信噪比的局部数据;最后,为提高微地震信号初至拾取效率,对传统AIC算法进行数学推导,将原始计算公式转换求取离散实数序列的算术和、平方和的线性组合形式,减少重复计算,得到快速AIC算法(Fast-AIC算法)。通过实际数据试算,当数据长度超过6500个采样点时,与传统AIC方法相比其计算效率提高超过1000倍。基于论文研究的技术方案,编程实现了基于Fast-AIC算法的微地震事件初至拾取及自动识别方法。采用合成数据对本文方法处理过程以及详细步骤进行论述,并进行大量实验分析本文方法的可靠性和准确性。同时,采用本文方法和几种常规方法对中国山西某地实际水力压裂微地震数据进行初至自动拾取,与人工拾取结果对比,本文方法拾取结果绝对误差更小。
[Abstract]:Microseismic monitoring technology is one of the most reliable methods for evaluating fracturing effect in oil field at present. Through accurate location of microseismic events, fracture strike, dimension and inversion mechanism can be judged, which provides a basis for subsequent production and development of oil field. Automatic identification of microseismic events and first break pick-up is one of the most important techniques. The picking speed and precision of seismic wave at first arrival directly affect the efficiency of micro-seismic positioning and the reliability of positioning results. Micro-seismic monitoring methods are mainly used in Tianjin. The micro-seismic source locates several hundred meters or even several kilometers below the earth's surface, and the signal-to-noise ratio of micro-seismic records is generally low, which leads to the limitations of the existing methods of automatic identification of micro-seismic events and first break picking in processing actual micro-seismic data. In order to accurately pick up the first arrival time of microseismic signals corresponding to all microseismic events, the number and time of microseismic events induced by hydraulic fracturing are uncertain. Firstly, the number of microseismic events and the approximate induced time are determined, and then the local microseismic data are picked up precisely according to the induced time of microseismic events. Akaike Information Criteria (AIC) method is simple to implement and easy to calculate. It is especially suitable for local data acquisition with microseismic events. However, the calculation efficiency of large-scale microseismic data processing for long-term records needs to be improved. In order to overcome the shortcomings in the above studies, a new method based on Fast-AIC algorithm is proposed to pick up the first break of microseismic signals. In order to overcome the dependence of micro-seismic identification method based on perforation signal on pre-perforation data, a layered velocity model based on acoustic logging curve is proposed and ray tracing theory is used to forward modeling when there is no actual perforation data. The perforation signal is calculated instead of the actual perforation signal. Then, according to the geophone arrangement parameters and formation velocity parameters of hydraulic fracturing monitoring system, the local effective data is selected near the trigger time of micro-seismic events, and the local data is filtered by Curvelet transform to provide high signal-to-noise ratio for accurately picking up micro-seismic signals. Finally, in order to improve the efficiency of picking up the first break of microseismic signals, the traditional AIC algorithm is deduced mathematically, and the original formula is transformed to get the arithmetic sum of discrete real number sequence and the linear combination form of square sum. The fast AIC algorithm (Fast-AIC algorithm) is obtained by reducing the repeated calculation. Compared with the traditional AIC method, the computational efficiency is more than 1000 times higher after 6500 sampling points. Based on the technical scheme studied in this paper, the method of picking up the first arrival of microseismic events and automatic identification based on Fast-AIC algorithm is programmed and realized. The reliability and accuracy of this method are tested and analyzed. At the same time, this method and several conventional methods are used to automatically pick up the first break of the actual hydraulic fracturing microseismic data in Shanxi, China. Compared with the results of manual picking, the absolute error of the picking result is smaller.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TE357.1;P631.4

【相似文献】