耦合生成对抗和自编码神经网络重构三维多尺度多孔介质

流程：用GAN估计出粗结构，再根据GAN结果用AE估计出细结构。输入数据，用的是二维数据CT数据 ，重建的是三维数据。

Coupled generative adversarial and auto-encoder neural networks to reconstruct three-dimensional multi-scale porous media

Highlights
•The networks have been used to reconstruct realizations of 3D porous media.
网络被用来重建3D多孔介质

•The multi-scale reconstruction of a Sandy microstructure is done.
完成砂质微结构的多尺度重建

•Three-dimensional coarse-images (low-resolution images) are predicted by GAN.
通过GAN预测三维粗图像(低分辨率图像)

•The coarse-image is converted to fine-image (high-resolution images) using AE.
使用AE将粗图像转换为细图像(高分辨率图像)。

•Porosity, permeability, auto-correlation function and visual inspection are evaluated for reliability of proposed methods.
评估了孔隙度、渗透率、自相关函数和目视检验的可靠性。

在本研究中，我们使用耦合产生式对抗与自动编码器神经网路来重建三维多孔介质的实现。采用基于梯度下降的优化方法训练和稳定神经网络。对伊朗某油田的砂岩和碳酸盐岩样品进行了多尺度重建。砂岩含有粒间和粒内孔隙。生成对抗网络预测粒间孔隙，自动编码器提供具有粒内孔隙(微孔隙)的生成对抗网络结果。不同的匹配标准，包括孔隙度、渗透率、自相关函数和目视解释，被用来研究模型的性能。该方法为研究人员重建多孔介质的多尺度实现提供了一种可靠的方法。

In this study, coupled Generative Adversarial and Auto-Encoder neural networks have been used to reconstruct realizations of three-dimensional porous media. The gradient-descent-based optimization method is used for training and stabilizing the neural networks. The multi-scale reconstruction has been conducted for both sandstone and carbonate samples from an Iranian oilfield. The sandstone contains inter and intra-grain porosity. The generative adversarial network predicts the inter-grain pores and the auto-encoder provides the generative adversarial network result with intra-grain pores (micro-porosity). Different matching criteria, including porosity, permeability, auto-correlation function, and visual interpretation have been used to investigate the performance of the models. This methodology provides researchers with a reliable method to reconstruct multi-scale realizations of porous media.

[1]. Shams, R., et al., Coupled generative adversarial and auto-encoder neural networks to reconstruct three-dimensional multi-scale porous media. Journal of Petroleum Science and Engineering, 2020. 186: p. 106794.【三区】