The Use of 3D Convolutional Autoencoder in Fault and Fracture Network Characterization

Conventional pattern recognition methods directly use 1D poststack data or 2D prestack data for the statistical pattern recognition of fault and fracture network, thereby ignoring the spatial structure information in 3D seismic data. As a result, the generated fault and fracture network is not distinguishable and has poor continuity. In this paper, a fault and fracture network characterization method based on 3D convolutional autoencoder is proposed. First, in the autoencoder training frame, 3D prestack data are used as input, and the 3D convolution operation is used to mine the spatial structure information to the maximum and gradually reduce the spatial dimension of the input. Then, the residual network is used to recover the input’s details and the corresponding spatial dimension. Lastly, the hidden features extracted by the encoders are recognized via - means, SOM, and two-step clustering analysis. The validity of the method is verified by testing the seismic simulation data and applying real seismic data. The 3D convolution can directly process the seismic data and maximize the prestack texture attributes and spatial structure information provided by 3D seismic data without dimensionality reduction and other preprocessing operations. The interleaving convolution layer and residual block overcome low learning and accuracy rates due to the deepening of networks.

[1]  Treatment and Effect of Loess Metro Tunnel under Surrounding Pressure and Water Immersion Environment , 2020 .

[2]  Hui Liu,et al.  Analysis of deformation characteristics and stability mechanisms of typical landslide mass based on the field monitoring in the Three Gorges Reservoir, China , 2018, Journal of Earth System Science.

[3]  M. Matos,et al.  Unsupervised seismic facies analysis using wavelet transform and self-organizing maps , 2007 .

[4]  Yujing Jiang,et al.  Feasibility investigation of the mechanical behavior of methane hydrate-bearing specimens using the multiple failure method , 2019, Journal of Natural Gas Science and Engineering.

[5]  Haibin Di,et al.  Non-linear GLCM texture analysis for improved seismic facies interpretation , 2017 .

[6]  Guangmin Hu,et al.  Unsupervised seismic facies analysis via deep convolutional autoencoders , 2018 .

[7]  Zhen Zhang,et al.  Stability analysis of a typical landslide mass in the Three Gorges Reservoir under varying reservoir water levels , 2020, Environmental Earth Sciences.

[8]  Yong-gang Zhang,et al.  A novel dynamic predictive method of water inrush from coal floor based on gated recurrent unit model , 2020, Natural Hazards.

[9]  Yanghua Wang.  Reservoir characterization based on seismic spectral variations , 2012 .

[10]  Vikram Jayaram,et al.  A comparison of classification techniques for seismic facies recognition , 2015 .

[11]  Guangmin Hu,et al.  Multi-waveform classification for seismic facies analysis , 2017, Comput. Geosci..

[12]  S. Geiger,et al.  Multiscale fracture network characterization and impact on flow: A case study on the Latemar carbonate platform , 2015 .

[13]  P. Yan,et al.  Discrete element numerical simulation of mechanical properties of methane hydrate-bearing specimen considering deposit angles , 2020 .

[14]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  Diganta Misra,et al.  Mish: A Self Regularized Non-Monotonic Neural Activation Function , 2019, ArXiv.

[16]  Junkun Tan,et al.  The influence of water level fluctuation on the stability of landslide in the Three Gorges Reservoir , 2020, Arabian Journal of Geosciences.

[17]  Abhinav Gupta,et al.  Non-local Neural Networks , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[18]  Robert M. Haralick,et al.  Textural Features for Image Classification , 1973, IEEE Trans. Syst. Man Cybern..

[19]  Yong-gang Zhang,et al.  Application of an enhanced BP neural network model with water cycle algorithm on landslide prediction , 2020, Stochastic Environmental Research and Risk Assessment.

[20]  Daniel Zoran,et al.  Pooling is neither necessary nor sufficient for appropriate deformation stability in CNNs , 2018, 1804.04438.

[21]  Yong-gang Zhang,et al.  A novel displacement prediction method using gated recurrent unit model with time series analysis in the Erdaohe landslide , 2020, Natural Hazards.

[22]  H. Mohammadzadeh,et al.  Geofluids Assessment of the Ayub and Shafa Hot Springs in Kopet-Dagh Zone (NE Iran): An Isotopic Geochemistry Approach , 2017 .

[23]  Guangmin Hu,et al.  Prestack Reflection Pattern Based Seismic Facies Analysis , 2015 .

[24]  D. Sanderson,et al.  The use of topology in fracture network characterization , 2015 .