基于FCM的微地震初至自动拾取算法研究

发布时间：2018-03-29 17:01

本文选题：微地震勘探　切入点：初至自动拾取　出处：《吉林大学》2017年硕士论文

【摘要】：利用水力压裂技术对低渗透型油气藏压裂注水时会引发变化的地下应力场,致使地下岩层错动开裂而诱发微地震事件。目前国内外微地震监测技术是处理水力压裂最为重要的手段,主要是通过在压裂井附近排布三分量检波器,对微地震有效事件进行连续记录,从而实现开断裂缝和油气藏位置的实时监测。微地震监测技术以其独有的优势能够对地下裂缝和压裂效果进行合理的解释与评价,进一步为利用水力压裂技术调整改造低渗透型油气田提供理论依据和技术支持。微地震监测技术需要实时、自动的对检波器接收的高密度微地震勘探资料进行解释,实现震源的高精度定位。而微地震初至拾取是实现高精度定位震源的重要环节,在保证微地震记录精度的基础上,设计实时快速自动稳定的初至拾取算法,是提高微地震资料处理速度与野外监测分析效率的关键。实际微地震资料有效信号的主要特点是频率高、能量弱、持续时间短,且受复杂强随机噪声的严重干扰,导致信噪比极低。人工拾取微地震初至效率低且不能适应实际生产的实时处理需求。传统自动拾取微地震初至算法如能量比算法、Akaike信息准则(AIC)算法等,在拾取高信噪比微地震记录初至时有一定的效果,但是在微地震资料质量较差时,其拾取结果难以满意。本文从微地震有效信号与复杂噪声的特征差异入手,分析了微地震随机干扰的类型和特性,并对有效信号的时间特性进行了全面系统的研究,并以此为基础提出了基于模糊C-均值聚类(FCM)的微地震初至自动拾取算法。该算法以微地震有效信号与噪声的特征差异为理论基础,通过不断优化目标函数得到能够表征数据相似性程度的隶属度矩阵。隶属度矩阵的值大于某一预先设定的阈值时,其值对应的样本点被划分为有效信号类,认为微地震数据的初至时刻是有效信号类的起始时刻。此外本文还给出了所选模糊指标和隶属度矩阵初始化参数的数值大小和原因,并详细论证了所选参数的可行性和有效性。本文全面分析了FCM算法的拾取性能以及模拟与实际微地震资料的拾取效果。通过绘制受试者工作特征(ROC)曲线与误差统计直方图验证了本文算法拾取初至的可靠性与准确性,最低能够处理信噪比为-8d B的微地震数据,且准确率高达86%。将该算法应用于模拟与实际微地震资料初至拾取,并与能量比算法和AIC算法进行对比,对比结果表明本文算法拾取结果的准确率优于其他两种算法,解决了传统初至拾取算法在低信噪比时难以有效拾取微地震初至的难题。
[Abstract]:Using hydraulic fracturing technology, the underground stress field will change when fracturing water injection in low permeability reservoirs. At present, microseismic monitoring technology is the most important means to deal with hydraulic fracturing, mainly by arranging three-component geophone near fracturing wells. The effective events of microearthquakes are recorded continuously, so as to realize the real-time monitoring of fracture breaking and oil and gas reservoir location. The microseismic monitoring technology can explain and evaluate the underground fractures and fracturing effect reasonably with its unique advantages. It provides theoretical basis and technical support for the adjustment and reconstruction of low permeability oil and gas fields by hydraulic fracturing technology. Micro-seismic monitoring technology needs to interpret the high-density microseismic exploration data received by geophone in real time and automatically. To realize the high precision location of the earthquake source, and the detection of the first arrival of the micro-earthquake is an important link to realize the high-precision location of the earthquake source. On the basis of guaranteeing the precision of the micro-seismic record, a real-time, fast, automatic and stable algorithm of the first arrival pickup is designed. It is the key to improve the processing speed of microseismic data and the efficiency of field monitoring and analysis. The main characteristics of the effective signal of practical microseismic data are high frequency, weak energy, short duration and serious interference by complex and strong random noise. As a result, the signal-to-noise ratio (SNR) is extremely low. The initial arrival efficiency of manual pickup microearthquakes is low and can not meet the real time processing requirements. The traditional automatic pickup micro-seismic first arrival algorithm, such as the energy ratio algorithm and Akaike information criterion (AICI) algorithm, is used in this paper. There is a certain effect in picking up high signal-to-noise ratio (SNR) microseismic records, but when the quality of microseismic data is poor, it is difficult to obtain satisfactory results. This paper starts with the difference of the characteristics between the effective signals of micro-earthquakes and the complex noises. The types and characteristics of random interference of microearthquakes are analyzed, and the time characteristics of effective signals are studied systematically. Based on this algorithm, a fuzzy C- mean clustering algorithm for automatic pickup of microearthquakes is proposed. The algorithm is based on the differences between the characteristics of effective signals and noises of micro-earthquakes. Through continuous optimization of the objective function, a membership matrix can be obtained, which can represent the degree of similarity of the data. When the value of the membership matrix is larger than a predetermined threshold, the sample points corresponding to the values are divided into effective signal classes. It is considered that the initial arrival time of the microseismic data is the starting point of the effective signal class. In addition, the numerical value and reason of the initialization parameters of the selected fuzzy index and membership matrix are also given in this paper. The feasibility and validity of the selected parameters are demonstrated in detail. In this paper, the picking performance of the FCM algorithm and the pickup effect of simulated and actual microseismic data are comprehensively analyzed. The histogram verifies the reliability and accuracy of the algorithm. Microseismic data with a signal-to-noise ratio of -8dB can be processed at the minimum, and the accuracy is as high as 86. The algorithm is applied to the first arrival pickup of simulated and actual microseismic data, and compared with the energy ratio algorithm and the AIC algorithm. The comparison results show that the accuracy of the proposed algorithm is better than that of the other two algorithms, which solves the problem that the traditional first arrival pick-up algorithm is difficult to effectively pick up the initial arrival of micro-earthquakes at low SNR.
【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】