Intelligent Generation of Cross Sections Using a Conditional Generative Adversarial Network and Application to Regional 3D Geological Modeling

The cross section is the basic data for building 3D geological models. It inefficient to draw a large number of sections build an accurate model. This paper reports use multi-source and heterogeneous data, such as maps, gravity aeromagnetic by conditional generative adversarial network (CGAN) implements intelligent generation method overcome problem modeling based on CGAN. Intelligent are carried out in three different areas Liaoning Province. results show that: (a) accuracy proposed higher than GAN Variational AutoEncoder (VAE) models, achieving 87%, 45% 68%, respectively; (b) model constructed generated our study consistent with manual creation terms stratum continuity thickness. suggests that significant surmounting difficulty processing involved regional modeling.