Gravity anomaly inversion method and system based on convolutional neural network, terminal and medium

Abstract

The invention belongs to the technical field of geophysical inversion, and discloses a gravity anomaly inversion method based on a convolutional neural network, and the method comprises the steps: constructing two-dimensional density models of different shapes, carrying out the forward modeling to obtain gravity anomaly data, and constructing a sample data set based on the obtained gravity anomaly data; constructing a gravity anomaly inversion network, namely AlexNet-Gra; preprocessing and dividing the constructed sample data set; utilizing the processed sample data set to train, optimize and verify the constructed AlexNet-Gra network, and obtaining a trained AlexNet-Gra network; and inputting the gravity abnormal data into the trained AlexNet-Gra network, so that an inversion result can be obtained. According to the method, after a large amount of data is obtained through forward modeling by means of models designed into different shapes, the position and density of the gravity anomaly body can be accurately inverted, and the gravity anomaly inversion problem can be effectively solved.

Description

technical field [0001] The invention belongs to the technical field of geophysical inversion, and in particular relates to a gravity anomaly inversion method, system, terminal and medium based on a convolutional neural network. Background technique [0002] At present, gravity exploration is one of the very important exploration methods in geophysical exploration. It has the advantages of wide detection area, low cost and wide working field. Fields such as surveys have wide-ranging applications. Gravity exploration measures the distribution of the earth's gravity field, studies the gravity anomalies produced by geological bodies with uneven density, performs targeted data processing, and completes corresponding geological interpretation tasks. It is a key step in gravity exploration and the most important step in gravity inversion to establish a corresponding model by using the characteristics of anomaly distribution, and to interpret the gravity anomaly quantitatively acco...

AI Technical Summary

Problems solved by technology

[0005] (1) The existing technology is easy to fall into a local minimum, the calculation is time-consuming, and it will occupy a large amount of computer memory when the amount of data is large

[0006] (2) The amount of data in the existing technology is small, which cannot meet the needs of a large amount of magnetic anomaly data required when using deep learning to solve geophysical inversion problems

[0007] (3) Most of the existing technologies rely on the initial model; the inversion of existing technologies is prone to problems such as underfitting or overfitting

[0011] (3) Since the gravity anomaly data does not have a fixed depth resolution, for the samples of the vertical double model, it is easy to have a glue phenomenon, resulting in poor inversion results

Images