.png){kind=icon}
.png){kind=icon}
Return
A 3-D Magnetotelluric Inversion Method Based on the Joint Data-Driven and Physics-Driven Deep Learning Technology
.png){kind=icon}
PRE
AI
DOI：10.1109/TGRS.2024.3369179.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The conventional magnetotelluric (MT) inversion method is subject to the influence of the initial model, which leads to an unstable inversion process and a tendency to get trapped at local optima. In contrast, deep learning technology relies on its powerful nonlinear fitting capability and can construct complex nonlinear mappings directly from observation data (input) to model (output). In recent years, it has received extensive attention from researchers. Due to the difficulties in creating a sufficiently large dataset and performing extensive neural network training, most current MT inversion methods for geophysical exploration remain limited to 1-D or 2-D scenarios. To the best of our knowledge, for deep learning-based 3-D MT inversion, currently there is no reported work in the literature. In this work, we propose a 3-D MT inversion method based on deep learning technology. By designing a neural network architecture for 3-D structures (MT3D-Net), we achieve an end-to-end mapping from the network input to output. To alleviate the excessive dependence of the network on the training set, we introduce a joint weighted loss function based on data-driven and physics-driven method, allowing the network to follow the physical constraints of MT data during the training process and thus more reasonably guide the update of network parameters. Numerical experiments show that this method combines the advantages of the traditional and data-driven inversions, significantly improving the stability and accuracy of MT inversion. The proposed method has been successfully applied to synthetic models and measured field data, and it has good application prospects.
Keywords:
Three-dimensional displays
Magnetic domains
Training
Neural networks
Data models
Solid modeling
Deep learning
3-D inversion
data-driven
deep learning
magnetotelluric (MT) method
physical constraints
Journal
IF:
8.6
Papers:
2.1W
Citations:
10.7W
View more organizations