Two-dimensional Transient electromagnetic data based on U-NET

Transient electromagnetic method is an active source time-domain electromagnetic method that achieves geological exploration by observing the decay process of the secondary electromagnetic field generated by an ungrounded closed loop or a grounded long straight conductor. In recent years, machine learning has achieved good results in various fields, and this paper applies the U-NET network in machine learning to the two-dimensional inversion of transient electromagnetic method, developing a U-NET-based two-dimensional inversion algorithm for transient electromagnetic method. Different models are designed for verification, validating and discussing the feasibility of convolutional neural networks in geophysical exploration.