Reservoir model multi-scale fine characterization method based on concurrent generative adversarial network

The invention provides a reservoir model multi-scale fine characterization method based on a concurrent generative adversarial network. Multi-scale representation is carried out on an original training image through a pyramid structure, and representation results of different scales keep a consistent spatial distribution mode. Under the condition that a receptive field is fixed, corresponding global information and local information can be extracted from a small-scale image and a large-scale image respectively. Meanwhile, a concurrent training strategy is applied, random initialization is replaced by a parameter inheritance mode between adjacent stages, and it can be ensured that parameters of the network model are fully and effectively trained. And finally, carrying out difference measurement on the generated distribution and the real distribution by utilizing a Wasserstein distance, limiting the gradient of the loss function within a certain range by adopting a gradient penalty strategy, ensuring the stabil