Application Of Kalman Filtering In The Magnetic Full-tensor Gradiometer Data Processing

Mineral resources are important strategic resources of national economy, so detecting their reserves and distribution is an important work. And in the detection of geological mineral resources, geophysical methods are important means. Magnetic prospectingplays an important role in the regional tectonic inference, the classification of magnetic rock mass, andthe mapping of rock formation as a kind of geophysical methods. With the development of modern information technologies, instruments and computer technologies, aerospace magnetic prospecting gradually becomes the mainstream of magnetic prospecting.This paper studies deeply inthe basic principles of the full tensor magnetic gradient measurement system, and completes the procedure of forward modeling based on the theoretical research. And then it analyzes the sources of noise in the measuring process,and in this paper, we select different filter methods based on the types of noise to do oppressive experiments. One-dimensional profile data processing mainly utilizes the median filter and kalman filter, and two-dimensional grid data mainly adopts median filter, gaussian low-pass filter and kalman filter, whichall take kalman filter as the main research target. Through the selectionof disciplinary functionsbased on the model tests, we improveits application in the whole magnetic gradient tensor data.The tests of theoretical models show that using kalman filter can make noise suppressed effectively, and obtain data with higher signal-to-noise ratio. In practical application, we choose the experimental data of Shanghai Hengsha Island’s full tensor magnetic gradient, and the method realizes the basic application in the actual data processing.Based on the above research contents, this paper mainly achieves the following results:(1) Studying the forward modeling of whole magnetic gradient tensor data, and coding to achieve the basic forward modeling, which lays the foundation for the subsequent pretreatment work on data;(2) According to the characteristics of full tensor magnetic gradient measurement, we studythe method of the existing median filter, gaussian low-pass filter and kalman filter, and code to realize the corresponding filters, then prove the applicability of those filter methods by model data;(3) Researching on the basic theory of kalman filter, we need master its basic properties and applications, and establish a model to verify its selection on parameterscompared to other filter methods, which can get some related research results;(4) Based on one-dimensional and two-dimensional data filter methods, programmingcan realize visualization on data processing;(5) Getting the relevant conclusions by the comparison on processing results of different filters for full tensor magnetic gradient data.