Three-dimensional magnetic tensor gradient inversion method and device based on convolutional neural network

Abstract

High-precision vector magnetic field detection is widely applied to the fields of celestial body magnetic field detection, aeromagnetic detection, ocean magnetic field detection, geomagnetic navigation and the like at present and is installed on airborne magnetic detection platforms such as satellites, airplanes, unmanned aerial vehicles, submarines and the like, and due to the fact that the data acquisition amount is large, calculation-intensive processes such as processing of a large amount of magnetic vector data and three-dimensional inversion are involved. The invention provides a novel inversion method for directly extracting parameters from a magnetic tensor gradient data image by using a CNN (Convolutional Neural Network) and synthesizing the parameters to generate a model matched with a detection target body. The method comprises the following steps of: performing forward modeling on a synthetic source body to obtain enough magnetic tensor gradient data samples; and adjusting a volume CNN structure, adding a shear layer, and realizing prediction of each parameter. Through single and double cube model numerical simulation, the algorithm accuracy is verified. According to numerical simulation and comparison test results, the CNN network has good nonlinear inversion capability, can realize accurate and rapid prediction of magnetic tensor gradient inversion, and is high in practicability and wide in application range.

Description

technical field [0001] The invention relates to the field of magnetic tensor gradient inversion, in particular to a three-dimensional magnetic tensor gradient inversion method and device based on a convolutional neural network. Background technique [0002] Tensor gradient measurement is an important detection method in magnetic detection, especially full tensor measurement, which not only preserves the intensity and direction information of the vector, but also ensures that the common mode suppression field from the earth's core, from deep in the earth's crust. Regional fields and geomagnetic changes from the ionosphere and magnetosphere, as well as higher resolution, can reveal detailed features such as superimposed anomalies. It has the advantages of strong anti-interference ability and high precision, and is widely used in the fields of celestial magnetic field detection, aeromagnetic detection, ocean magnetic field detection and geomagnetic navigation, etc. It is instal...

AI Technical Summary

Problems solved by technology

Neural networks have achieved some success in geophysical inversion, but almost no application in magnetic field tensor gradient inversion

Images