Abstract:Fault identification plays a vital role in geological structure interpretation, reservoir characterization, and reservoir evaluation. The traditional process for fault identification is label intensive. Deep learning-based, automatic fault detection from seismic data has attracted extensive attention in recent years. However, using a 2D neural network (i.e., single seismic data slice) leads to the loss of correlation between seismic slices, and using a 3D neural network (i.e., 3D seismic data cube) increases the workload of fault labeling and the difficulty of training. To overcome the drawbacks of 2D and 3D neural networks, in this work, we introduce a channel attention mechanism in the original U-net architecture. The new model, which is referred to as 2.5D channel attention U-net (2.5D CAU-net), takes four adjacent seismic slices as inputs instead of the entire 3D seismic data cube. The efficacy of 2.5D CAU-Net for fault detection is demonstrated using publicly available synthetic seismic datasets and real seismic field datasets. We explored the influence of cropping the original seismic data into various sizes on fault identification. Experimental results show that the 2.5D CAU-net can effectively utilize the correlation information among adjacent seismic slices to improve fault detection performance. For different seismic data cropping strategies, results indicate that the more extensive size cropping method allows the network to obtain more contextual information and more complete fault features, improving the efficacy of fault picking. However, the cropping approach with a larger size leads to a smaller number of labeled examples. When the amount of training data is not large enough, the larger size cropping approach tends to lead to model over-fitting, which will decrease the performance of the fault recognition, especially in a 3D neural network. Overall, 2.5D CAU-net can improve the efficiency of fault recognition with less annotation effort and computational resources.

Fault Identification of U-Net Based on Enhanced Feature Fusion and Attention Mechanism

Fault Recognition Method Based on Attention Mechanism and the 3D-UNet

A method for intelligent identification of faults in seismic using an attention-based ES-UNet network with model re-training learning

Automatic seismic fault identification based on an improved U-Net network

Automatic identification of seismic faults via integrating Residual Network-50 residual blocks and convolutional block attention modules

Automatic Identification of Seismic Faults Via the Integration of ResNet-50 Residual Blocks and Convolutional Attention Modules

Automatic Geological Fault Identification from Seismic Data Using 2.5D Channel Attention U-net

Automatic Geologic Fault Identification from Seismic Data Using 2.5D Channel Attention U-net

Fault Identification Based on W-Net in 3-D Seismic Images

U_Net & Residual Neural Networks for Seismic Fault Interpretation

3-D Seismic Fault Detection Using Recurrent Convolutional Neural Networks With Compound Loss

Automatic 3D fault segmentation based on multi-scale feature fusion model with compound loss function

MultiURNet for 3D seismic fault attributes fusion detection combined with PCA

An Intelligent Recognition Method for Low-Grade Fault Based on Attention Mechanism and Encoder–Decoder Network Structure

Transformer assisted dual U-net for seismic fault detection

Common-azimuth Seismic Data Fault Analysis Using Residual UNet

CNN Based Fault Recognition with Multi-Scale Fusion Attention Mechanism

Enhanced Seismic Data Segmentation Using an Assembled Scse-Res-unet Deep Neural Network

Research on Fault Recognition Method Combining 3D Res-UNet and Knowledge Distillation

Fault3DNnet: A Lightweight 3D Seismic Fault Detection Network with Bidirectional Decoding

Autoencoder-assisted Feature Ensemble Net for Incipient Faults