Tranform-based magnetotelluric inversion model establishment method

The invention discloses a magnetotelluric inversion model establishing method based on transform. The magnetotelluric inversion model establishing method comprises the steps of collecting a magnetotelluric sounding model sample data set; performing forward modeling conversion by adopting a Maxwell equation to obtain data of apparent resistivity and phase samples corresponding to a sample data set of the magnetotelluric sounding model, and forming sample pairs; splitting the sample pair into a training set and a test set; the method comprises the following steps: establishing a SwinUnet network based on a Transform architecture, and carrying out learning mapping; training the established SwinUnet network by using the training set to obtain an optimal inversion network weight and a fitting nonlinear function obtained by fitting the hyper-parametric model with the apparent resistivity and the data of the phase sample, and completing the training of the SwinUnet network; testing the trained SwinUnet network by us