小波降噪在核磁共振测井中的应用研究

发布时间：2018-03-21 00:34

本文选题：核磁测井　切入点：回波串　出处：《华中科技大学》2015年硕士论文　论文类型：学位论文

【摘要】：降低核磁共振测井信号中的噪声是提高测井信噪比的一项基本任务。被动降噪一般处于核磁共振测井回波串形成之后的数字信号处理阶段。相关应用资料表明,小波降噪算法是一种消除数据噪声的有效方法。本文在介绍核磁共振测井原理的基础上,对测井信号及其回波串的构成机制进行了必要的阐述,并分析了噪声在测量系统中产生的机理、存在形式及其对测量主要参量精度的影响。论文介绍了小波降噪算法的数学基础和小波降噪算法的构成,选用了两种小波降噪算法:Mallat算法和Lifting Scheme(提升框架)算法,利用两种算法对多组不同信噪比(10-30dB)的数据进行降噪处理,并对相关性能指标进行了分析,仿真结果表明,Lifting Scheme算法在整体性能指标上略优于Mallat算法。另外,应用测井工程中用于标定的一组柴油数据作为参照标准,分别对模拟窄带滤波方法,Mallat算法,Lifting Scheme算法的降噪效果进行对比实验测试,结果表明两种小波算法均优于窄带滤波方法并且Lifting Scheme算法要优于Mallat算法。论文还分别对所选两种算法的现场可编程门阵列(Field-Programmable Gate Array,FPGA)设计实现问题进行了分析探讨,在计算量相同的条件下,Lifting Scheme算法在计算速度、硬件资源占用量等方面也优于Mallat算法,从而从算法性能及其在FPGA设计实现两方面验证了Lifting Scheme算法在核磁共振测井系统中应用的可行性。
[Abstract]:Reducing noise in nuclear magnetic resonance logging signal is a basic task to improve signal-to-noise ratio. Passive noise reduction is generally in the digital signal processing stage after the formation of nuclear magnetic resonance logging echo string. Wavelet denoising algorithm is an effective method to eliminate data noise. Based on the introduction of nuclear magnetic resonance (NMR) logging principle, the formation mechanism of logging signal and its echo string is expounded. The mechanism of noise generation in measurement system, its existing form and its influence on the accuracy of main measurement parameters are analyzed. The mathematical basis of wavelet denoising algorithm and the constitution of wavelet denoising algorithm are introduced in this paper. Two kinds of wavelet denoising algorithms: Mallat algorithm and Lifting Schema algorithm are selected. The two algorithms are used to deal with multiple groups of data with different signal-to-noise ratio (SNR) of 10-30dB, and the related performance indexes are analyzed. The simulation results show that lifting Scheme algorithm is slightly better than Mallat algorithm in overall performance index. In addition, a set of diesel oil data used for calibration in logging engineering is used as a reference standard. The effect of noise reduction in the proposed lifting Scheme algorithm, an analog narrowband filtering method, is tested by a comparative experiment. The results show that both wavelet algorithms are superior to narrow band filtering methods and Lifting Scheme algorithm is superior to Mallat algorithm. The design and implementation of field programmable gate array Field-Programmable Gate FPGA are also discussed in this paper. The lifting Scheme algorithm is better than the Mallat algorithm in computing speed and the amount of hardware resources, etc. Thus, the feasibility of Lifting Scheme algorithm in NMR logging system is verified from the performance of the algorithm and its design and implementation in FPGA.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN791

【参考文献】