Earthquake earth surface deformation analysis method and device based on deep learning and Beidou positioning

The invention provides a seismic earth surface deformation analysis method and device based on deep learning and Beidou positioning, and relates to the technical field of seismic earth surface deformation data analysis, and the method comprises the steps: obtaining the original observation data of each Beidou seismic station in a target region, and constructing a to-be-processed coordinate time sequence; through the first neural network model, performing gross error elimination processing and missing value interpolation processing on the to-be-processed coordinate time sequence to obtain a target coordinate time sequence; fitting parameter values of a pre-constructed Beidou seismic station speed calculation model based on the target coordinate time sequence; and training the second neural network by using the speed result of each Beidou seismic station and the station feature information to obtain an earth surface deformation analysis model. According to the method, automatic processing of gross error elimina