摘要
摘要
石油和天然气素有工业血液之称，是一个国家的重要战略性资源。近年随着未勘探的易开发区域逐渐减少，地球物理研究人员已将油气勘探的重点目标转向对技术要求更高的隐藏性油气藏和复杂构造性油气藏。目前单纯的储层弹性或物性参数反演已不能完全满足油气勘探的需要，而利用叠前地震资料开展高分辨率的储层弹性与物性参数同步反演已成为学术界和工业界共同关注的热点问题。
本文在回顾地震资料高分辨率处理和地震反演的研究背景及意义的基础上，总结归纳出了目前叠前地震反演中存在的两个问题：1）如何获得高分辨率的地震资料；2）如何实现高分辨率的叠前地震同步反演的确定性优化方法。为解决以上两个问题，本文分别做了如下创新工作：
1、针对问题1），本文提出了一种基于BP人工神经网络的地震资料高分辨率处理方法。该方法利用BP人工神经网络建立井旁地震道记录振幅谱与补偿系数之间的非线性映射关系，进而利用该关系计算出其它待补偿地震记录振幅谱的补偿系数，接着对补偿系数进行空间加权平滑和自适应补偿位置选择处理，最后将其作用于振幅谱，得到补偿后的高分辨率地震记录。该方法相较于其它方法，同时采用了测井资料信息和地震资料信息，尽量避免了补偿不足或补偿过多的现象，增强了补偿依据。
2、针对问题2），本文提出了一种基于双参数弹性速度模型的储层弹性与物性参数叠前地震同步反演的确定性优化方法。该方法利用双参数弹性速度模型建立储层弹性与物性参数之间的联系，在贝叶斯反演框架下以储层弹性与物性参数联合的后验概率为目标函数，同时利用地震资料高分辨处理方法提升同步反演初值的分辨率，最后用自适应变步长优化方法求解得到分辨率更高的储层弹性和物性参数。一方面本方法采用双参数弹性速度模型，与其它岩石物理模型相比，该模型能够更好的建立弹性参数与孔隙形状参数之间的联系，有助于认识孔隙形状对储层弹性性质的影响；另一方面本方法采用确定性优化方法构建反演框架和求解，与随机优化方法相比，反演速度更快、精度更高。
本文提出的方法均运用实际工区数据进行了验证，从验证效果来看，本文提出的地震资料高分辨率处理方法在提升地震资料分辨率上有明显的效果；本文提出的同步反演的确定性优化方法具有很好的收敛速度和稳态效果，井曲线投上去后吻合度好，满足叠前反演要求。
关键词：高分辨率，神经网络，叠前同步反演，贝叶斯框架，自适应变步长
ABSTRACT
Oil and natural gas is long known as the industrial blood, which is an important strategic resource of a country and related to the economic and social development. In recent years, as easy to develop unexplored areas gradually reduce, geological exploration rearchers have taken the focus of the oi
l and natural gas exploration target to the hidden rervoir and complex rervoir construction, tho technical requirements are higher. So just rely on the physical or petrophysical parameters of rervoir could not fully meet the needs of oil and natural gas exploration, and using the pre-stack ismic data to carry out high resolution ismic joint inversion methods for estimating physical and petrophysical parameters of rervoir has become the hot topic in academia and industry.
Bad on the review of the background and significance of high resolution ismic data processing and ismic inversion, we summarized out two unanswered questions of the pre-stack ismic inversion:1) how to get the high resolution ismic data; 2) how to realize the high resolution pre-stack ismic joint inversion with deterministic optimization method. In order to solve above two problems, this paper did the specific works of the innovation as follows:
1. For the question 1), this paper propod a new high resolution ismic data processing method bad on BP artificial neural network method. Firstly, we ud BP artificial neural network to establish the mapping relationship between the amplitude spectrum of the practical ismic information of nearby the well and the compensation coefficient. Secondly, we ud the relationship to calculate each compensation coefficient of amplitude spectrum of ismic records without compensation. Thirdly, we procesd the compensation coefficient with weighted smoothing and ada
ptive compensation location lection method respectively. Finally it acted on the amplitude spectrum to get the high resolution ismic records. This method takes the spectrum width of logging data as the compensation standard, which overcomes the shortcomings of tho general spectrum methods which lack of quantitative compensation standard and enhance the compensation basis.
2. For the question 2), this paper propod a new method for estimating physical and petrophysical parameters of rervoir bad on double parameters elastic velocity model with deterministic ismic pre-stack inversion. This method ud double
parameters elastic velocity model to build up the relationship between pre-stack data and petrophysical parameters We treated the joint posterior probability of physical and petrophysical parameters as the objective function under the Bayesian architecture, and by using the high resolution ismic data processing methods to improve the resolution of the initial value of joint inversion. we ud the adaptive variable step length optimization method to get higher resolution physical and petrophysical parameters of rervoir in the end. On the one hand, this method adopts double parameters elastic velocity model, compared with the other rock physical model, this model can better establish the relationship between physical parameters and the pore shape parameter, he诀别的意思是什么
lp to know the influence of pore shape on the rervoir elastic properties. On the other hand, this method adopts the deterministic optimization method to establish the inversion framework and solve it, compared with the stochastic optimization method, it has faster inversion speed and higher precision.
The propod methods were tested by using the real data respectively, from the testing results, the new high resolution ismic data processing method has the obvious effect in raising the resolution of ismic data. The new joint inversion of deterministic optimization method has good convergence speed and steady effect, and the inversion results fit the well logging curve well. It proved that this method conform to the requirements of the pre-stack inversion.
inboxKeywords:high resolution, artificial neural network, pre-stack joint inversion, Bayesian architecture, adaptive variable step length optimization method
目录
第一章绪论 (1)
1.1研究背景和意义 (1)
1.2 国内外研究现状 (2)
1.2.1 地震资料高分辨率处理研究现状 (2)
1.2.2 地震反演方法研究现状 (3)
1.3 本文主要工作与贡献 (6)
paradigm1.4 本文组织结构 (7)
第二章叠前地震反演理论基础 (8)
2.1 叠前地震反演概述 (8)
2.2 A VO反演基本原理及方法 (8)
2.2.1 Zoeppritz及其近似方程 (9)
2.2.2 正演模型的建立 (12)
2.2.3 A VO反演方法 (14)
2.3 叠前地震反演面临的问题 (16)
2.3.1 提升地震观测资料的分辨率 (17)
2.3.2 实现叠前同步反演的确定性方法 (18)沈阳日语翻译
2.4 本章小结 (18)
第三章基于BP人工神经网络的地震资料高分辨率处理方法 (20)
3.1 引言 (20)
niece
3.2 数据预处理 (21)
3.2.1 地震子波的确定 (21)
3.2.2 生成合成地震记录 (23)
3.2.3 时频变换 (23)
3.3 基于BP人工神经网络的地震资料高分辨率处理方法 (24)
3.3.1 BP人工神经网络结构的确定 (24)
3.3.2 补偿系数求解与空间加权平滑处理 (28)
3.3.3 自适应补偿处理 (30)
英语四级答题技巧3.3.4 地震记录的恢复 (30)
3.4 实验仿真及分析 (31)
3.5 本章小结 (39)
办公用品的英文第四章高分辨率的储层弹性与物性参数同步反演方法 (40)
怎样有效去除色斑4.1 引言 (40)
4.2 储层弹性与物性参数同步确定性反演方法 (40)
4.2.1 双参数弹性速度模型 (41)
touristy
4.2.2 同步反演方法框架 (45)
4.2.3 自适应变步长优化方法 (49)
4.2.4 同步反演的高分辨率处理 (50)
4.3 实验仿真与分析 (52)
4.3.1 合成数据测试 (52)
4.3.2 实际数据应用 (55)
4.4 本章小结 (60)
第五章总结与展望 (61)
5.1 工作总结 (61)
5.2 工作展望 (62)
致谢 (63)
参考文献 (64)
个人简历 (68)
攻读硕士学位期间取得的研究成果 (69)
攻读硕士学位期间参加的科研项目 (69)
>role是什么意思
