【作者】 崔建军；

【导师】 何继善； 金尚柱；

【作者基本信息】 中南大学 ， 地球探测与信息技术， 2001， 博士

【摘要】 多年的勘探表明，我国石油地质条件复杂，油气埋深大，隐蔽性强，油气资源勘探正面对越来越困难的条件。对于地震勘探而言，就要求它能够提供更多、更详细和更精确的信息。复杂构造成像一直是地震勘探关注的焦点问题，实现地下复杂构造的准确成像，对于提高油气勘探的成功率、降低勘探风险具有极高的应用价值和实际意义，所以研究复杂构造成像的理论方法和实现技术成为当前亟待解决的问题。 经典弹性数学理论是对精确遵守虎克定律的理想介质建立的，而实际大地介质是具有粘滞性的粘弹性介质，所以完全弹性波动方程不能够准确地描述地震波在实际介质中的传播过程。地震波在大地介质的传播过程中，不仅波的振幅会衰减，而且有频散现象，因此要从地震数据中得到复杂构造的准确信息和高分辨率的图像，就必须研究粘弹性介质中波的传播规律。修改后的Stokes粘弹性波动方程所描述的介质相当于开尔芬体，可以利用它研究地震波在大地中的传播过程。本文从Stokes粘弹性波动方程出发，着重讨论粘弹性介质中地震波的衰减和频散特性，得到了地震波的低频成份在传播过程中，频散现象微弱可以忽略不计的重要结论。通过傅氏变换和复数运算，首次成功推导了均匀粘弹性介质中Stokes粘弹性波动方程的正演和偏移算法。以此为基础，本文首次提出能够适应于速度纵、横向变化的Stokes粘弹性波动方程叠后偏移算法和Stokes粘弹性波动方程叠前深度偏移算法，并分别进行了数值模拟。针对有些情况下将实际介质看做标准线性体来研究粘弹性介质中波的传播问题，本文从频率域中的线性粘弹性波动方程出发，推导了线性粘弹性波动方程的偏移算法，通常用复数速度来计算吸收问题，但是复数速度是无法测量的，本文简单介绍了利用品质因数确定复数速度的方法。 建立速度模型是复杂构造成像的基础，而速度参数反演问题本身是非线性的。近年来发展起来的神经网络技术具有解决任意复杂非线性问题的能力，本文对BP神经网络算法进行了改进，并提出基于神经网络技术的速度分析方法，该方法可以分别用于叠前域和偏移后域的速度分析，对噪声有很强的适应能力。偏移速度分析方法具有精度高，可以唯一确定速度场分布等优点，基于本文建立的粘弹性波动方程偏移算法，本文研究了基于粘弹性波动方程的偏移速度分析方法。 摘耍 多学科、多种方法相结合的原则己经成为复杂构造成像的指导思想，本文针对先验信息这一成像的重要前提条件，首次提出了基于己知先验信息的复杂构造成像方案和基于未知先验信息的复杂构造成像方案。复杂构造成像是大计算量的数据处理过程，需要密集型的计算技术，这一直限制着精确成像技术的应用，据此，本文介绍了一种基于Agent技术的分布式计算方法，并首次在地震数据处理领域提出基于Agent的数据处理实现技术，该项技术具有智能性和灵活性，可以充分地利用Internet和Intranet上的机器资源，实现快速有效的数据处理，本文设计了整体的实现方案。更多还原

【Abstract】 The exploration of these years shows that the oil geologic condition of China is very complicated, the depth of oil and gas is deep and concealed. The situation of oil and gas resource exploration becomes more and more difficult. Facing such a hard situation, seismic exploration is required to enclose more rigorous information with details. Complex structure imaging is a focus in the field of seismic exploration lately. Imaging the complex structure underground earns great application significance and practical value for increasing the hitting rate of target and decreasing the risk of exploration. The research work on theory and realizing techniques of complex structure imaging has been a hard problem to solve urgently.The classical elasticity mathematics theory was established for ideal medium abiding Hooke law grimly, the earth medium is a viscoelastic medium with viscosity in fact, so elastic wave equation can not be used to describe the propagation of seismic waves in real media. The viscosity of the medium will cause dissipation of seismic energy, thus decreasing the amplitude and modifying the frequency content of the propagating wavelet. In order to obtain accurate information and imaging with high distinguishability from seismic data, the propagating rule of wave in viscoelastic medium should be studied. The medium described with modified Stokes viscoelastic wave equation is a Kelvin solid, which can be used to study the propagating of seismic wave in earth media. The characteristics of attenuation and dispersion of seismic wave in viscoelastic medium is discussed with emphasize on the base of Stocks viscoelastic wave equation. An important conclusion that the weak dispersion of low frequency component of seismic wave can be ignored. Through the Fourier transformation and complex operation, the basic algorithm of forward and migration of Stokes viscoelastic wave euqation in uniform viscoelastic medium is deduced successfully for the first time. On this basis, poststack migration algorithm and prestack depth migration algorithm of Stokes viscoelastic wave equation are put forward to fit velocity variety both in vertical and horizontal, and the numerical simulation are pursued separately. In some research, the earth medium was regarded as standard linear solid to study the propagating rule of wave in viscoelastic medium. From this view, the migration algorithm of linearIllviscoelastic wave equation is deduced in frequency field. The absorption is computedby complex velocity, but it is impossible to measure. A resolution to acquire thecomplex velocity with quality factor is described in this paper also.The basis of complex structure imaging is establishing velocity model, thevelocity inversion is a nonlinear problem. The neuraI network developed in these yearshas a characteristic of solving nonlinear mapping with extremity complexity. The back-propag8tion algorithm of neural network is modified, and the velocity analyzingmethod based on neuraI network is investigated, which can be used on .velocityanalyzing in both prestack field and migration fieId. This method shows a strongadaptability to noise. The migration velocity analysis has many advantages such asprecision and fixing on velocity distributing exclusively. The migration velocityanaIysis based on viscoelastic wave equation is developed in this paper.The principle of comprehensive utiliZaion of multi-course and different methodshas been a guiding line in complex structure imaging. For apriori information being animportant precondition, imaging plans based on aPriori infOrmation and nonaprioriinformation are put forward separate1y. The complex structure imaging is a dataprocessing with a great number of computation and need high density computingtechnology, which limits the improvement of imaging. To break such a limit, adistributing computation based on Agent is brought forward in the fina1 part of thispaper, the technique of data processing based on Agent is induced in seismi更多还原