中国科学院软件研究所机构知识库
Advanced  
ISCAS OpenIR  > 中科院软件所  > 中科院软件所
题名:
沉积相综合分析方法研究及软件研制
作者: 孟庆武
答辩日期: 2001
专业: 计算机软件与理论
授予单位: 中国科学院软件研究所
授予地点: 中国科学院软件研究所
学位: 博士
关键词: 分维微分 ; 模糊判别 ; 推理框架 ; 属性聚类 ; VC理论 ; SVM学习 ; 神经网络 ; 三层ANN容量 ; 二次型映射
摘要: 对地下的研究难于对天体的研究,石油行业对高技术的依赖远超过其它一些行业。在石油勘探研究中沉积相的预测占重要地位,尤其是沉积微相预测,对于恢复古地理,确定沉积环境,研究地层的空间变化,分析油气的生、储、运、聚都具有重要意义。限于目前的测量手段和技术水平,要做好地下地层沉积相的预测是很难的。石油勘探具有多学科的性质,沉积相的分析要求综合利用地震、地质及测井数据,进行各种分析和处理,对地下地层沉积相的空间展布做出预测。目前在国内外都还没有系统的沉积相综合分析软件。在部分国外软件中嵌有一些沉积相分析的内容,但无论从功能上,还是方法上都离实际应用上的要求差的较远。实际勘探中的沉积相划分以人工分析为主,一般依靠经验,判别标准也各不相同,相的划分因人而异,差别较大。本论文的研究工作按照中科院软件所与大庆油田公司联合培养计划要求,结合石油勘探中存在的实际难点、重点问题,进行深入的理论和算法研究,同时研制开发较为实用的沉积相综合分析软件系统。充分利用和融合地震、地质、测井等所有来源的各种信息,进行全方位的、多方面的特征分析,减少人为因素影响,初步实现计算机自动划相,为油气藏评价、储量预测、井位部署提供依据。完成的工作主要有以下四个方面:1。地震资料的高分辨率处理技术 石油勘探中,由于成本极高,所钻的探井通常只有稀疏的几口,所以可用的岩心和测井等直接资料通常很少。地震资料是在地表测量的间接资料,得到的是反映地下信息的地层结构的三维数据体,提高其精度是整个石油勘探的基础。我们根据地震波传播机理,利用分形分维理论将地震与声波测井资料紧密结合起来,研制开发了一种地震资料的高分辨率处理技术,恢复由于地层吸收等作用而衰减的高频成分,反映地下真实结构,提高了地层分辨能力。2。地质模糊推理技术沉积相的分析和判定需要丰富的专业化知识,还要在多种知识之间作出选择。很多预测是不完全和不确定的。不同来源、不同学科的数据和知识对同一地层得出的结论可能相互佐证也可能截然不同。实际勘探中的人工划相依靠经验,判别标准也各不相同,相的划分因人而异。我们利用模糊技术实现综合地质推理。构造沉积相数据结构及知识库。建立不同沉积类型的地质、测井及地震相模式,统一判别标准。用属性框架表示沉积相态模式空间,建立模糊推理机制,对有限输入的地质、测井及地震属性特征,通过地质属性框架的模糊匹配进行综合推理分析,确定相态类型。3。地震属性聚类技术三维数据体地震资料包含了关于地下地层的大量有用信息,充分利用它们可以进行石油勘探的各种分析和预测。从地震资料中提取的各种属性参数多达三十多种。并不是所有属性参数都要用到,也并不是属性参数可以直接有来做各种预测。我们利用90年代后期提出并得到迅速发展应用的VC理论和SVM算法进行各种地震性参数分析。开发地震属性聚类技术,将属性参数向高维空间投影分离,并进行模式聚类,降低其VC维,解决沉积相预测的可解性问题,并借助地质理地质理论——沃尔相率解决训练数据的有效性问题,满足大样本要求。4。空间外推预测沉积相空间展布预测要在井约束下由多种地震属性参数外推得到。我们从理论上研究神经网络算法用于相态空间分布预测的可行性,数学上证明了三层神经网络隐层节点数与VC维的关系,为实际应用中隐层节点的选择提供了标准和理论依据。提出可变结构的改进神经网络算法,并用沉积相分布的综合预测。在传统的BP网络算法中引入搜索技术,在数据学习的过程中不断指导搜索方向,使单向传播的网络运行模型具有了一定的反馈功能,根据误差增大及减小趋势控制网络拓扑结构的复杂程度变化。使网络结构随所模拟模式不同而实时变化。
英文摘要: The research of underground is harder than that of celestial bodies, so the petroleum industry heavily depends on high technique than others. It is very important to predict the distribution of sedimentary facies, especially tiny sedimentary sub-facies in the research of petroleum exploration and helpful of ancient physiognomy rebuild, sedimentary condition identify, distribution research of strata, evolving, storing, carrying and gathering analysis of petroleum. Subjected to actual measure method and technical level, it is difficult to predict the distribution of sedimentary facies of underground strata by rule and line. As containing many fields of study in petroleum exploration, it is necessary to carry through various analysis and process in order to foretell facies distribution of underground strata by use of seismic data, geological data and well logs data together. There is to nitegrated analysis software of sedimentary facies at present both at home and overseas. Limited functions are contained in some foreign software. They are deficient both in function and method practically. Identify of sedimentary facies is mainly by manual work in actual exploration. There are great differences between different people in analysis as different experiences and criterions. According to the associated training scheme of Institute of Software of Chinese Academy of Sciences and Daqing Oil Field Corporation, this dissertation deals deeply with several important and difficult problems existed in actual petroleum exploration. Some new algorithms are presented and integrated analysis software of sedimentary facies is developed. Various data is fully used in character analysis from different sources, such as seismic data, geological data and well logs data. Artificial effects are decreased and realize automatically identify of sedimentary facies preliminarily. They provide bases for evaluation of oil and gas reservoir, prediction of reserves and design of well position. The following is the main research works: 1. High resolution process technique of seismic data Due to the high costs, usually only several wells are drilled during petroleum exploration, so few direct information is available, such as well core and log Seismic data are indirect data measured on the earth's surfacies, but we can get data body of three dimensions about underground information and strata structure. The precision improvement of seismic data is base of entire works of petroleum exploration. According to seismice wave transmission theory, this dissertation uses fractal method to combine seismic with sonic log. A technique of high-resolution process is developed to recover high frequency of seismic that has been attenuated by strata absorption. Processed seismic data can show the real underground structure in detail, and enhance the resolution and increasing the signal-noise ratio of seismic data. 2. Fuzzy reasoning technique of geology Special knowledge is needed in analysis and differentiating of sedimentary facies, and makes choice between several kinds of knowledge. Many predictions are imperfect and unsure. Conclusions to the same stratum from different sources and fields might support each other or in opposite. Different people usually get different results as manual analysis of sedimentary facies depending on experience with different criterion in actual exploration. This dissertation realizes integrated reasoning of geology with fuzzy technique. Data structure of sedimentary facies and repository are constructed. Geologic, well log and seismic patterns are determined with different kinds of sediment under uniform differentiating standard. Patterns space of sediment forms is denoted by attributes frame. Sedimentary facies is predicted with input attribute characters of geological, seismic data and well log though fuzzy reasoning. Fuzzy matching of geological attributes frame is processed in the entire patterns space with the most possibility principle. 3. Clustering technique of seismic attributes Large useful information is contained in three dimensions body of seismic data about underground strata. Various analysis and predictions are made with them in petroleum exploration. Up to thirty parameters of attribute can be calculated from seismic data. Not all parameters are used in a certain prediction and nor can directly used. VC theory and Support Vector Machine are presented in the end of nineties and got rapid development and applications. They are used in parameters analysis of seismic attribues in this dissertation. Clustering technique of seismic attributes is developed to project seismic attribute parameters in high dimensions space and class them with patterns. Solvability of sedimentary facies prediction can be solved by decrease the VC dimensions. Geological theory-Wolve facies rule is used to meet the need of large number of samples and the validity of training data. 4. Prediction algorithm of distribution Distributing prediction of sedimentary facies can be calculated by some algorithms from multi-parameters of seismic attributes under the restrictive conditions of wells. The feasibility of neural network algorithm is studied in theory to forecast the facies distribution of underground strata in this dissertation. It is also proved in math the relation between VC dimension and the number of hidden layer nodes. This provides criterion and theory base for the selection of hidden node number in actual application. Improved algorithm of neural network is presented in this dissertation with changeable structure. It is used in integrated prediction of sedimentary facies. Search technique is introduced in traditional BP algorithm. The direction of search is guided during the process of data training. This make the running model of one-way propagation have some feedback function. The complexity of neural network structure is controlled by the tendency of error change. The structure of neural network is real time changed according to the patterns of simulated relation.
语种: 中文
内容类型: 学位论文
URI标识: http://ir.iscas.ac.cn/handle/311060/6550
Appears in Collections:中科院软件所

Files in This Item:
File Name/ File Size Content Type Version Access License
LW004418.pdf(1168KB)----限制开放-- 联系获取全文

Recommended Citation:
孟庆武. 沉积相综合分析方法研究及软件研制[D]. 中国科学院软件研究所. 中国科学院软件研究所. 2001-01-01.
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[孟庆武]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[孟庆武]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.

 

 

Valid XHTML 1.0!
Copyright © 2007-2017  中国科学院软件研究所 - Feedback
Powered by CSpace