【作者】 李晋；

【导师】 汤井田；

【作者基本信息】 中南大学 ， 地质资源与地质工程， 2012， 博士

【摘要】 大地电磁强干扰分离技术一直是大地电磁测深领域的研究热点和难点。迄今为止,它的研究工作已经取得了许多成果,但随着人类文明的不断发展,重工业密集等因素造成的环境噪声以及人类活动等因素造成的人文电磁噪声日益严重,导致大地电磁测深数据受到严重污染,大地电磁测深工作面临巨大困难。现有的大地电磁强干扰分离方法在矿集区实际应用和测试中表现出诸多不足,这一领域面临的困难和挑战也日益加剧。因此,为了提高大地电磁测深数据质量,抑制噪声干扰已成为当务之急。研究大地电磁强干扰的特征,提出有针对性的大地电磁强干扰分离方法,对改善大地电磁测深数据质量以及对大地电磁法探测结果的处理和解释具有重要意义。本文正是在这一背景下,在国家科技专项“深部矿产资源立体探测及实验研究”(SinoProbe-03)和国家自然科学基金“基于数学形态学的大地电磁信号与强干扰分离方法研究”(41104071)的联合资助下,利用数学形态学理论对大地电磁强干扰分离方法进行了深入研究,具有重要的理论和实际意义。本文基于数学形态学的思想,对大地电磁强干扰分离及应用展开了分析,重点研究了传统形态滤波和广义形态滤波的大地电磁强干扰分离方法以及在形态滤波的基础上,研究了Top-hat变换、中值滤波和信号子空间增强的大地电磁二次信噪分离方法。论文通过理论分析、模拟仿真以及实际应用等手段,围绕数学形态学开展大地电磁强干扰分离的研究工作。本文工作的主要贡献和创新总结如下：(1)研究了五种典型的大地电磁强干扰类型的特征规律,分析了矿集区主要的噪声来源。对一类点分别添加类方波干扰和类充放电三角波干扰,从时间域波形和卡尼亚电阻率测深曲线两方面研究了典型噪声干扰对大地电磁数据质量的影响情况。(2)数值模拟了典型的单一噪声干扰,研究了不同类型结构元素及尺寸的去噪性能,讨论了结构元素长度及类型的选取规律。(3)针对V5-2000不直接提供读取时间序列的软件,剖析了该仪器的数据采集格式,实现了大地电磁原始资料的读取及还原。提出了基于传统形态滤波的大地电磁信噪分离方法,分析了不同类型结构元素及同一类型、不同尺寸结构元素的去噪性能。(4)构建了组合广义形态滤波器,提出了基于组合广义形态滤波的大地电磁强干扰分离方法。在青海柴达木盆地开展了相关试验研究,选取了具有一定代表性的试验点进行组合广义形态滤波处理。对比了时间域波形和卡尼亚电阻率-相位测深曲线的改善情况,分析了该方法对包含比较单一的噪声干扰测点的去噪效果。对矿集区强干扰测点进行了组合广义形态滤波处理,综合评价了该方法对包含复杂噪声干扰类型的强干扰测点的噪声抑制能力。(5)在数学形态滤波的基础上,提出了基于Top-hat变换、中值滤波和信号子空间增强的大地电磁二次信噪分离方法。针对形态滤波提取的噪声轮廓或重构信号,进一步分离出包含大尺度低频细节成份的有用信号。对矿集区强干扰测点进行了二次信噪分离处理,对比分析了组合广义形态滤波和二次信噪分离方法的卡尼亚电阻率一相位测深曲线的改善情况,综合评价了两种方法在保留低频缓变化信息方面的优势,以及对大地电磁测深数据质量的改善效果。通过以上五个方面的研究表明,基于数学形态学的大地电磁强干扰分离方法有效地剔除了大地电磁强干扰中的大尺度干扰和基线漂移,较好地还原了大地电磁原始信号特征,改善了大地电磁测深数据质量。由于数学形态学运算速度快,具有潜在优势,为矿集区海量大地电磁信号与强干扰的分离提供了一条新的解决途径,应用前景广阔。最后,总结了全文的主要内容和创新点,讨论了数学形态学在大地电磁强干扰分离中的不足之处,并对下一步研究工作的开展提出了一些建议。更多还原

【Abstract】 The magnetotelluric strong interference separation technology has always been the research hot and difficulty. So far, it has made a lot of achievements, but with the continuous development of human civilization, environmental noise caused by heavy industry intensive factors and humanities electromagnetic noise caused by human activities factors are growing more and more serious, result in magnetotelluric sounding data are serious pollution and magnetotelluric faced enormous difficulties. Existed magnetotelluric strong interference separation methods show many deficiencies in the practical application and measure. The difficulties and challenges are also increasing in this field. Therefore, in order to improve the quality of magnetotelluric sounding data, suppress the noise interference has become imperative. Studied the magnetotelluric strong interference characteristics and proposed the specific method, are the important significance to improve the magnetotelluric sounding data quality, and the processing and interpretation of the magnetotelluric method detection results. This work are co-funded by the National Scientific and Technological Project of Deep Probing on3D Structure and Geodynamic Process of Ore District (SinoProbe-03) and the National Natural Science Foundation of China of the Research of Magnetotelluric Strong Interference Separation Method based on Mathematical Morphology (Grant No.41104071), and has important theoretical and practical significance using mathematical morphology theory study the magnetotelluric strong interference separation method.Based on the idea of mathematical morphology, we analyze the magnetotelluric strong interference separation and application, and focus on the traditional morphological filtering and the generalized morphological filtering as well as secondary signal-to-noise separation method of top-hat transformation, median filtering and signal subspace enhancement on the basis of morphological filtering. By means of theoretical analysis, simulation, and practical application, we carry out magnetotelluric strong interference separation research. The main contribution and innovation of this work are summarized as follows: (1) Study five typical magnetotelluric strong interference characteristics, and analyze the major noise sources in ore concentration area. Through add the similar square wave interference and the similar charge and discharge triangular wave interference to the measuring point, we study the quality of magnetotelluric data from both time-domain waveform and Cagniard resistivity curve impact on the typical noise.(2) Numerical simulate the typical single noise interference, and study the de-noising performance of different sizes and types of structural elements, moreover, discuss the selection rules of the structural elements sizes and types.(3) According to the V5-2000does not directly provide time series software, analyzing the instrument data acquisition format, realizing the magnetotelluric original material reading and restore. The work proposes the magnetotelluric signal-to-noise separation method based on traditional morphological filtering, and analyzes the de-noising performance of different type structural elements and the same type, different size structural elements.(4) The work constructs the combination generalized morphological filtering, and proposes magnetotelluric strong interference separation method based on the combination generalized morphological filtering. Qaidam basin in Qinghai Province, we carry out test research, and select a representative measuring point by using the combination generalized morphological filtering for processing. Compared with improvement situation both time domain waveform and Cagniard resistivity-phase curve, analyzed de-noising effect of measuring point including comparative single noise. Through the combination generalized morphological filtering to process the strong interference measuring point, comprehensive evaluated the noise suppression capability of strong interference measuring point including complex noise interference types.(5) On the basis of the mathematical morphological filtering, we propose magnetotelluric secondary signal-to-noise separation methods of top-hat transformation, median filtering and signal subspace enhancement. According to the noise contour or reconstructed signal extracted by morphological filtering, and further separated the useful signal which contains large-scale low frequency detail components. Using the secondary signal-to-noise separation to process the strong interference measuring point in ore concentrated area, we comparative analyze the Cagniard resistivity-phase curve improvement situation both the combination generalized morphological filtering and secondary signal-to-noise separation method, and comprehensive evaluate the advantages of the two methods on the reservation of low frequency slow change information, as well as the quality improvement effect for magnetotelluric sounding data.Through the above five aspects research, we show that, magnetotelluric strong interference separation method based on mathematical morphology can effectively eliminate large-scale interference and baseline drift for magnetotelluric strong interference, better restore the magnetotelluric original signal characteristics and improve the quality of magnetotelluric sounding data. Due to the mathematical morphology operation speed is fast, and it has potential advantages. The method provides a new solution for the separation of the mass magnetotelluric signals and strong interference in ore concentrated areas, and has broad application prospect.Finally, we summarize the main contents and innovations of this work, and discuss the deficiencies of mathematical morphology in the magnetotelluric strong interference separation, moreover, put forward some suggestions on the next stage of research work.更多还原