回归分析在分离叠加磁异常中的应用研究

发布时间：2019-02-22 10:36

【摘要】：本文是从最佳线性滤波理论出发,对目前磁测资料数据处理中分离区域场与局部场的两种滤波—匹配滤波和维纳滤波的频率响应特性作了分析比较,并指出匹配滤波只是一般维纳滤波的一个特例。在信息化技术高度发展的21世纪,电子计算技术迅速发展,在国内外重磁资料数据处理中,频率域滤波技术的应用日益增多。众所周知,不管是在航空磁测还是在我们的地面磁测工作中,我们所测量得到的磁异常数据均是所探测对象的一个综合反映,即是不同埋深磁性地质体异常叠加的综合体现。而深源场的频谱主要集中于低频段,浅源场则具有很强的中、高频成分。根据频谱的这种特征差异,我们就可以将反映深源的区域场与反映浅源的局部场加以分离。首先,根据回归分析的应用示例建立相应的回归分析模型,通过对模型的实验分析,得出可以把在对叠加异常滤波处理中遇到的非线性回归问题转换为线性回归问题来决解的结论,并且所建立的回归分析模型是有效可行的。然后,通过对球体和二度水平圆柱体正演模型磁场的平面特征及剖面特征进行讨论分析。获得单个规则磁性异常体的平面等值线和主剖面曲线,并分析了主剖面曲线的特征点(如极值点、拐点、零值点等)与磁异常体中心埋深、中心在地面投影点以及有效磁化倾角的对应关系。在此基础之上,进一步讨论了不同埋深的磁性球体叠加模型、二度水平圆柱体叠加模型以及球体-二度水平圆柱体叠加模型的平面等值线和主剖面曲线的特征,对比分析了单一模型体和叠加组合模型体的异常曲线特征。对正演得到的叠加组合模型的△T主剖面异常曲线进行两种不同方法的滤波处理。一是,根据△T主剖面叠加异常曲线直接对其使用维纳滤波或者匹配滤波方法进行滤波处理;第二种滤波方法就是在滤波处理的过程中引入回归分析模型。通过对两种方法的对比分析知道,引入回归分析方法的滤波方法对叠加场源的分离效果更好。最后,将此方法应用于工区实测资料中,对工区实测剖面进行了异常的分离处理,同样得到了很好的效果,说明回归分析方法对叠加磁异常滤波分离有一定的辅助作用。
[Abstract]:In this paper, based on the theory of optimal linear filtering, the frequency response characteristics of the two kinds of filtering, matching filtering and Wiener filtering in the data processing of the magnetic measurement data are compared and compared, and the matching filtering is only a special case of general Wiener filtering. In the 21st century of the height development of the information technology, the electronic computing technology has developed rapidly, and the application of frequency domain filtering technology is increasing in the data processing of the heavy magnetic data at home and abroad. It is well known that the magnetic anomaly data measured by our measurement is an integrated reflection of the detected object, that is, the comprehensive reflection of the abnormal superposition of different buried deep magnetic geological bodies, whether in the aeromagnetic measurement or in our ground magnetic measurement. and the frequency spectrum of the deep source field is mainly concentrated in the low frequency section, and the shallow source field has strong middle and high frequency components. Depending on this feature of the spectrum, we can separate the field from the deep source and the local field that reflects the shallow source. First, according to the application example of the regression analysis, a corresponding regression analysis model is established, and by the experimental analysis of the model, the conclusion that the non-linear regression problem encountered in the superposition abnormal filtering process can be converted into a linear regression problem can be obtained, and the established regression analysis model is effective and feasible. Then, the plane characteristic and the profile characteristic of the magnetic field of the spherical and the second-degree horizontal cylinder are analyzed and analyzed. The plane contour line and the main section curve of a single regular magnetic anomaly body are obtained, and the corresponding relation between the characteristic point of the main section curve (such as the extreme point, the inflection point, the zero value point, etc.) and the center of the magnetic anomaly body, the center of the center on the ground projection point and the effective magnetization inclination angle are analyzed. on the basis of which, the characteristics of the plane contour line and the main section curve of the different buried deep magnetic sphere superposition model, the second-degree horizontal cylinder superposition model and the sphere-two-degree horizontal cylinder superposition model are further discussed, The characteristics of the abnormal curves of a single model and a superimposed composite model are compared and analyzed. and two different methods of filtering processing are carried out on the abnormal curve of the static T main section of the superposed combined model which is being played. In one embodiment, a Wiener filter or a matched filtering method is directly used for filtering processing according to the superposition of the abnormal curve of the main section of the IGT; the second filtering method is to introduce a regression analysis model in the process of filtering processing. Through the comparative analysis of the two methods, it is known that the filtering method introduced to the regression analysis method has better separation effect on the superposition field source. In the end, this method is applied to the measured data of the work area, and the abnormal separation and treatment are carried out on the measured section of the work area. The results show that the regression analysis method has a certain auxiliary effect on the superposition magnetic anomaly filtering and separation.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P631.2

【参考文献】