基于测井曲线的地层划分和测井相分析

发布时间：2018-05-02 16:28

本文选题：测井曲线 + 平滑滤波器　；参考：《西安科技大学》2015年硕士论文

【摘要】：随着测井综合解释的成熟,在地质解释中综合应用各种测井曲线已成为一种趋势。本文以测井曲线为研究对象,对地层划分和测井相分析中的关键方法进行研究与实现。主要研究成果和创新点如下:1.利用面向对象的解析方法,通过分析存储测井曲线的WIS文件,对测井曲线数据进行了正确的解析和存储。针对测井数据的影响因素,应用固定窗长相关对比法对曲线进行深度校正,应用Savitzky-Golay平滑滤波器滤除噪声,尽可能的消除非地层因素的影响,为测井曲线划分和测井相分析提供了正确的数据支持。2.根据油田取芯数据的小样本特性,结合同区域地质特征相似性,研究基于支持向量机的测井曲线分层。针对传统多分类算法在计算上的不便及未考虑地层连续特性,改进了一对一多分类方法,提高了样本的学习和预测速度。由于每一段地层为一些连续点,应用滤波函数和阶梯函数识别地层分界面,进一步提升了地层划分的准确性。3.结合同区域同层位内测井相的相似性,研究了基于测井曲线的支持向量机并用来进行测井相分析。能够反映测井相的测井曲线较多且这些数据之间存在相关性,因此应用主成分分析提取测井相主成分,降低了样本维数,有效地提升了计算效率。通过实验比较说明了编码多分类方法的有效性,应用基于主成分分析的支持向量机编码多分类方法对预测井进行了分析,得到了良好的分析结果。
[Abstract]:With the maturation of comprehensive log interpretation, it has become a trend to apply various logging curves in geological interpretation. In this paper, the key methods of formation division and logging phase analysis are studied and realized. The main research results and innovations are as follows: 1. By using the object oriented analysis method, the log curve data are correctly analyzed and stored by analyzing the WIS files that store the log curves. In view of the influence factors of logging data, the depth of the curve is corrected by using the correlation correlation method of fixed window length, and the noise is filtered by using Savitzky-Golay smoothing filter, and the influence of non-formation factors is eliminated as much as possible. It provides correct data support for log curve partition and log phase analysis. According to the small sample characteristics of the core data and the similarity of geological characteristics in the same region, the stratification of log curves based on support vector machine (SVM) is studied. In view of the inconvenient calculation of the traditional multi-classification algorithm and the failure to consider the continuous characteristics of the strata, the one-to-one multi-classification method is improved, and the learning and prediction speed of the samples is improved. Because each formation is some continuous points, the filter function and step function are used to identify the stratigraphic interface, which further improves the accuracy of stratigraphic division. The support vector machine (SVM) based on log curve is studied and used to analyze the phase of well logging by combining the similarity of the well facies in the same zone and the same layer. It can reflect the logging curves of logging facies and there is correlation between these data, so the principal component analysis is used to extract the principal components of logging facies, which reduces the sample dimension and effectively improves the calculation efficiency. The effectiveness of the coding multi-classification method is demonstrated by experimental comparison. The prediction well is analyzed by using the support vector machine coding multi-classification method based on principal component analysis, and a good analysis result is obtained.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.81;TP18

【相似文献】