探地雷达多次波识别及压制方法研究

发布时间：2018-02-12 02:35

本文关键词： 探地雷达自适应预测误差反褶积多次波识别多次波压制　出处：《中国地质大学(北京)》2017年硕士论文　论文类型：学位论文

【摘要】：探地雷达技术是一种非侵入浅表地球物理探测技术,它利用超高频脉冲电磁波(106-109Hz)来确定介质的分布规律。当存在介电常数相差较大的界面时,探地雷达的接收天线便会收集到多次波,常见的该类介质包括空洞、金属、管线、含水层、低速薄层等。目前探地雷达数据采集多选用剖面法,得到的数据仅能表现反射面的起伏变化,但数据中不含电磁波速度等信息。因此在识别多次波时,能够利用的信息很少,存在很大的识别难度。目前探地雷达剖面数据中多次波的压制多借鉴地震数据的处理方式,但雷达数据相比于地震数据又有很多不同,它的超高频、短波长以及采集方式的不同使得地震数据的处理方式不能完全应用于雷达数据。多次波的存在严重干扰着有效信息的识别,如果压制处理不完全还会损伤有效信息,造成解释陷阱。随着探地雷达技术的应用日益广泛,其获得数据中的多次波问题也越显突出,而目前关于探地雷达多次波的识别特征和压制方法的研究却很少,因此研究探地雷达多次波的识别及压制方法就变得更有意义。本文为研究探地雷达剖面中多次波的周期性和非周期性特征,针对地表与地下地层呈水平状态、地下反射层与地表倾向不一致等情况下,多次波所表现出的形态差别明显的问题,设计了多个不同的地质模型进行数值模拟,分析研究了数据中多次波的识别特征。同时针对整个剖面中非周期性多次波的的压制问题,提出了自适应预测误差反褶积技术。该技术在预测反褶积技术的基础上进行了改进,借助于LMS最小均方算法,在循环迭代的过程中更新预测滤波因子。该技术相比于只能设置单一预测步长一步求取预测滤波因子的预测反褶积方法,具有多次波预测准确,压制效果好的特点;相比于设置初始滤波因子为O的自适应反褶积技术,具有计算速度快、迭代次数少的特点。分别将该技术应用于正演数据中周期性和非周期性多次波的压制处理中,验证了该技术能够较好地压制层间多次波和自由表面多次波。之后将该技术应用于空洞模型物理实验数据和倾斜地层模型物理实验数据中,压制了数据中周期性和非周期性的自由表面多次波。最后在处理巴丹吉林沙漠沙丘内部结构探地雷达探测数据时应用了该技术,顺利完成数据中非周期性自由表面多次波的压制任务。
[Abstract]:Ground penetrating radar (GPR) is a non-invasive shallow geophysical detection technique, which uses ultra-high frequency pulsed electromagnetic waves 106-109 Hz to determine the distribution of media. The receiving antenna of the ground penetrating radar will collect multiple waves. The common types of media include holes, metals, pipelines, aquifers, low-speed thin layers, etc. At present, the profile method is often used for data collection of ground penetrating radars. The obtained data can only show the fluctuation of the reflection surface, but there is no information such as the velocity of the electromagnetic wave in the data, so there is very little information that can be used in the identification of multiple waves. At present, the suppression of multiple waves in ground penetrating radar profile data uses for reference the processing methods of seismic data, but the radar data is different from seismic data in many ways, and its ultra-high frequency (UHF). Because of the difference of short wavelength and acquisition method, seismic data processing method can not be applied to radar data completely. The existence of multiple waves seriously interferes with the recognition of effective information, and if the suppression processing is not complete, the effective information will be damaged. With the increasing application of ground penetrating radar (GPR) technology, the problem of multiple waves in the acquisition of GPR data becomes more and more prominent, but there are few researches on the identification characteristics and suppression methods of GPR multiples. In order to study the periodic and aperiodic characteristics of multiple waves in GPR profiles, this paper aims at the horizontal state of ground surface and underground strata. Under the condition that the underground reflection layer and the surface inclination are not consistent, many different geological models are designed for numerical simulation. The recognition characteristics of multiples in data are analyzed and studied. At the same time, an adaptive prediction error deconvolution technique is proposed for the suppression of non-periodic multiples in the whole section, which is improved on the basis of predictive deconvolution technology. With the aid of the LMS minimum mean square algorithm, the predictive filtering factor is updated in the process of cyclic iteration. Compared with the predictive deconvolution method, which can only be calculated by a single prediction step size, it has the accuracy of multiple prediction. Compared with the adaptive deconvolution technique which sets the initial filter factor O, it has a fast calculation speed. The technique is applied to the suppression of periodic and aperiodic multiples in forward data, respectively. It is proved that this technique can suppress interlayer multiple and free surface multiple well, and then it is applied to the experimental data of cavity model and inclined stratum model. The periodic and aperiodic free surface multiples in the data are suppressed. Finally, the technique is applied to the processing of the data from the ground penetrating radar (GPR) in the dunes of the Badanglin Desert. Successfully completes the data non-periodic free surface multiple suppression task.
【学位授予单位】：中国地质大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P631.3

【参考文献】