Two dimensional forward simulation of magnetotelluric waves based on fourier neural network

Abstract Magnetotelluric method is a geophysical exploration technology with large detection depth, high efficiency and low cost. Its theory, method and data processing and interpretation technology are relatively mature, so it has been increasingly widely applied in many industries in recent years. Accurate and efficient forward modeling is the basis for magnetotelluric method to be developed. Machine learning is a popular research topic in many fields. It can be used to simulate magnetotelluric two-dimensional forward modeling accurately and quickly, and Fourier neural network is a typical deep learning network in machine learning. In theory, after inputting the training model set, we use Fourier neural network to conduct network training, and the purpose of two-dimensional magnetotelluric forward modeling can be achieved quickly by using the trained network. In this paper, two-dimensional magnetotelluric forward modeling will be conducted based on Fourier neural network, and the generated neural network will be tested to verify whether the constructed network fits the training model and whether its forward prediction results are accurate.