Underground shallow seismic source positioning method based on deep learning

The invention relates to an underground shallow seismic source positioning method based on deep learning. The underground shallow seismic source positioning method comprises the following steps: arranging a distributed vibration sensor array, generating a learning sample, setting a seismic source bullet position corresponding to a three-dimensional energy field image sample as a training label, constructing a deep learning network framework, training a network, and positioning an actual explosion seismic source. According to the invention, the intermediate steps of positioning parameter extraction, positioning model modeling, positioning model calculation and the like in a traditional shallow seismic source positioning process are reduced. The method greatly improves the seismic source positioning efficiency, eliminates the positioning blind area, reduces the dependence of the channel reconstruction precision of a monitoring region on the seismic source positioning precision, and provides a new seismic source