基于互相关的全波形反演初始模型构建方法研究

发布时间：2018-03-25 10:41

本文选题：全波形反演　切入点：互相关　出处：《吉林大学》2017年硕士论文

【摘要】：在地震勘探数据的处理与解释中,速度是不可或缺的重要参数,对于地下介质构造的描述和储层预测都有重要作用。近年来油气田的勘探开发程度逐日提升,勘探目标从简单含油构造转向复杂构造及隐蔽油气藏,使得地震勘探对速度建模提出了更高的要求。传统的速度建模方法,例如走时层析成像和偏移速度分析等,往往不能满足复杂地质条件下地震资料处理解释所需的精度和分辨率。全波形反演是一种能够充分利用叠前地震波场全部信息的速度建模方法,它使用最优化方法对模拟记录和观测记录进行匹配,使两者之间的误差泛函最小化,从而得到相应的地下介质速度模型。全波形反演在复杂地质条件下也能较好地恢复出地下构造的细节,与传统速度建模方法相比具有更高的分辨率和精度。本文深入研究了二维时间域全波形反演的基本理论,推导了声波方程的规则网格高阶有限差分格式和目标函数的梯度公式,同时对全波形反演的稳定性进行了分析。由于常规全波形反演采用的2l范数形式的目标函数具有很强的非线性,加之模型参数变化复杂,使得反演过程不稳定,在迭代时容易陷入局部极小值。为了使反演更好地收敛到全局极小值,通常需要提供一个足够精确的初始速度模型或者需要地震资料包含较高信噪比的低频信息。针对这一问题,本文提出了一种基于互相关的初始模型构建方法,旨在为常规全波形反演提供一个较好的初始模型,以改善反演的稳定性。该方法使用互相关目标函数测量模拟记录与观测记录的相似性,侧重地震记录之间的相位匹配,具有更强的线性性质,因此更不容易陷入局部极小值。基于互相关目标函数的全波形反演能够恢复出模型的宏观构造,以此作为常规全波形反演的初始模型,可以有效提高最终反演结果的精度和稳定性。考虑到全波形反演的计算成本巨大,本文提出了一种时间域高效全波形反演方法,引入混合震源技术来提高计算效率。组成混合震源的各个独立震源之间相互影响会产生严重的串扰噪声,本文提出了一种动态随机编码方法来压制串扰噪声。该方法在每十次迭代后均重新随机抽取多个震源参与反演,并对各震源的振幅和相位进行随机编码。模型试验表明,基于动态随机编码的混合震源技术既能有效提高全波形反演的计算效率又能保证反演结果的质量。为了验证本文所提出的方法在地震数据缺少低频成分时的反演效果,使用高通滤波器滤除地震子波的低频成分,结果表明与常规方法相比,本文方法对低频信息的依赖性更低。另外还进行了噪声测试,进一步验证了本文所提出的方法在地震数据含噪声时也具有较好的反演效果和稳定性。
[Abstract]:In the process and interpretation of seismic exploration data, velocity is an indispensable parameter, which plays an important role in the description of underground media structure and reservoir prediction. In recent years, the degree of exploration and development of oil and gas fields has been increasing day by day. From simple oil-bearing structures to complex structures and subtle reservoirs, seismic exploration has put forward higher requirements for velocity modeling. Traditional velocity modeling methods, such as travel time tomography and migration velocity analysis, etc. The precision and resolution of seismic data processing and interpretation under complex geological conditions are often not satisfied. Full waveform inversion is a velocity modeling method which can make full use of all the information of prestack seismic wave field. It uses the optimization method to match the simulated and observed records to minimize the error functional between them. Thus the corresponding velocity model of underground media can be obtained. Full wave inversion can also recover the details of underground structure in complex geological conditions. Compared with the traditional velocity modeling method, this paper has higher resolution and accuracy. In this paper, the basic theory of 2-D time-domain full waveform inversion is studied, and the regular grid high-order finite difference scheme of acoustic equation and the gradient formula of objective function are derived. At the same time, the stability of full waveform inversion is analyzed. The inversion process is unstable because of the strong nonlinearity of the objective function in the form of 2l norm and the complex variation of the model parameters. It is easy to fall into local minima during iteration. In order to make the inversion converge better to the global minimum, It is usually necessary to provide a sufficiently accurate initial velocity model or to contain low frequency information with high signal-to-noise ratio (SNR) in seismic data. To solve this problem, a cross-correlation based initial model construction method is proposed in this paper. In order to provide a better initial model for conventional full-waveform inversion and improve the stability of inversion, this method uses cross-correlation objective function to measure the similarity between simulated and observed records, with emphasis on phase matching between seismic records. The whole waveform inversion based on the cross-correlation objective function can restore the macroscopic structure of the model, which can be used as the initial model of the conventional full-waveform inversion. It can effectively improve the accuracy and stability of the final inversion results. Considering the huge computation cost of the full waveform inversion, a time domain efficient full waveform inversion method is proposed in this paper. The hybrid source technology is introduced to improve the computational efficiency. The interaction between the individual sources that make up the mixed source will produce serious crosstalk noise. In this paper, a dynamic random coding method is proposed to suppress crosstalk noise. After every ten iterations, multiple sources are selected randomly to participate in the inversion, and the amplitude and phase of each source are randomly coded. The hybrid source technique based on dynamic random coding can effectively improve the computation efficiency of full waveform inversion and guarantee the quality of inversion results. In order to verify the inversion effect of the method proposed in this paper, when the seismic data is short of low frequency component, High-pass filter is used to filter the low-frequency components of seismic wavelet. The results show that the method is less dependent on low-frequency information than the conventional method. It is further verified that the proposed method has good inversion effect and stability when seismic data contain noise.
【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】：P631.4

【相似文献】