Geoacoustic Inversion Based on Neural Network

Traditional inversion methods, such as the matched field inversion, modal dispersion inversion, have been proposed and got good results. Still, the computing time of these methods is long due to large search space. With the development of artificial intelligence in recent years, deep learning methods have been utilized for geoacoustic inversion. An inversion framework based on neural networks is proposed in this work. The well-trained network can provide accurate inversion results when the processed data for inversion is given as the input of the neural network model. Additionally, the computational time will be shortened when using neural network to inverse geoacoustic parameters.