Hilbert-Huang变换在测井资料处理中的应用研究

发布时间：2018-01-24 17:03

本文关键词： Hilbert-Huang变换测井曲线去噪分析层序地层流体识别　出处：《西安石油大学》2017年硕士论文　论文类型：学位论文

【摘要】：目前,油气田地质研究依据的大多为常规测井资料。为了突破常规测井曲线“视觉域”分析的局限性,从常规测井数据中“挖掘”出更多的地质信息,本文将Hilbert-Huang变换应用于常规测井资料处理研究,以期为测井资料的处理解释提供新的方法,从而服务于测井地质研究。本文首先介绍了Hilbert-Huang变换的基本原理,重点阐述了EMD算法的分解过程、停止准则,总结了Hilbert-Huang变换的优缺点,并对仿真信号进行了分析。其次开展了基于Hilbert-Huang变换的测井曲线去噪分析,对比分析了HHT减性去噪、小波阈值去噪及HHT-WT联合去噪三种方法,并改变噪声的强度开展研究。基于Hilbert-Huang变换开展了层序地层划分,利用EMD分解得到的IMF分量较好地划分了层序界面及地层旋回。针对储层流体识别,对深感应电阻率曲线开展分析,利用得到的IMF2和Amp2的组合特征较好地识别了上油下水流体界面。最后对Hilbert-Huang变换在测井资料处理中的应用效果进行了分析总结。研究结果表明,Hilbert-Huang变换的核心部分是经验模态分解(EMD),该方法简单、分解效率高,具有自适应性;当噪声强度较小时,小波阈值去噪和HHT-WT联合去噪具有较好的效果,当噪声强度中等时,三种去噪方法均能获得较理想的结果,当噪声强度较大时,小波阈值去噪为首选方法;高频IMF可划分小层序(短期旋回),中频IMF可划分中等层序(中期旋回),低频IMF可划分大层序(长期旋回)。IMF分量相对于原始曲线,更加清晰地反映了层序界面及地层层序级别;油水界面处,IMF2分量达到峰值,Amp2分量达到阶段幅值最大。使用该方法时要注意剔除夹层对测井曲线的干扰,并注重多种测井系列的综合应用。
[Abstract]:At present, the geological research of oil and gas fields is mostly based on conventional logging data. In order to break through the limitations of "visual domain" analysis of conventional logging curves, more geological information is "mined" from conventional logging data. In this paper, Hilbert-Huang transform is applied to the study of conventional logging data processing, in order to provide a new method for processing and interpretation of logging data. In this paper, the basic principle of Hilbert-Huang transform is introduced, and the decomposition process and stop criterion of EMD algorithm are expounded. The advantages and disadvantages of Hilbert-Huang transform are summarized, and the simulation signal is analyzed. Secondly, the denoising analysis of logging curve based on Hilbert-Huang transform is carried out. The three methods of HHT de-noising, wavelet threshold de-noising and HHT-WT combined de-noising are compared and analyzed. And change the intensity of noise to carry out research. Based on Hilbert-Huang transform to develop sequence stratigraphy. The sequence interface and stratigraphic cycle are well divided by the IMF component obtained by EMD decomposition. The deep induction resistivity curve is analyzed for reservoir fluid identification. The combined features of IMF2 and Amp2 are used to identify the interface of oil and water fluid well. Finally, the application effect of Hilbert-Huang transform in logging data processing is analyzed. Summary. Research results show that. The core part of Hilbert-Huang transform is empirical mode decomposition (EMD), which is simple, efficient and adaptive. When the noise intensity is small, wavelet threshold denoising and HHT-WT combined denoising have a better effect. When the noise intensity is medium, the three denoising methods can obtain better results, when the noise intensity is high. Wavelet threshold denoising is the first choice; High frequency IMF can be divided into small sequence (short cycle cycle), intermediate frequency IMF can be divided into medium sequence (middle cycle) and low frequency IMF can be divided into large sequence (long term cycle. IMF component is relative to original curve). The sequence interface and stratigraphic sequence level are more clearly reflected; At the oil-water interface, the IMF2 component reaches the peak value and the Amp2 component reaches the maximum amplitude. When using this method, attention should be paid to eliminating the interference of intercalation to the logging curve and to the comprehensive application of various logging series.
【学位授予单位】：西安石油大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P618.13;P631.81

【相似文献】