Semi-supervised deep learning seismic data inversion method driven by wave equation

Abstract

This disclosure provides a semi-supervised deep learning seismic data inversion method driven by wave equations, which can realize the deep learning inversion network in the absence of corresponding geological models for some seismic data. The product-full connection network enhances the seismic data, and finally obtains the geological wave velocity model by extracting the characteristic map, and completes the mapping relationship between the seismic data and the underground multi-layer medium model; The seismic data of the geological model replaces the wave velocity loss function with a data loss function, introduces physical laws, and implements a semi-supervised learning strategy. Through the semi-supervised deep learning seismic data inversion network, the inversion effect of the deep learning network is improved when there are few labeled data.

Description

technical field [0001] The disclosure belongs to the field of geophysical exploration, and relates to a semi-supervised deep learning seismic data inversion method driven by wave equations. Background technique [0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art. [0003] Seismic method, as one of the most commonly used geophysical exploration methods, is widely used in petroleum exploration, coal field, and metal deposit detection, etc., and has broad application prospects. The main principle of the seismic method is based on wave field propagation. Multiple geophones are arranged on the surface, and the wave field is generated by stimulating the artificial seismic source multiple times and propagates in the underground medium. When the wave impedance of the underground medium changes, reflection or refraction returns to the ground. The geophones located on the ground re...

AI Technical Summary

Problems solved by technology

However, for seismic data, due to the difficulty in obtaining the information of the underground medium, the corresponding labels of all seismic data cannot be obtained, and the conditions for supervised deep learning cannot be met.

In addition, this data-driven algorithm lacks the support of physical meaning, and the generalization of the method is limited

Images