Research On Localization Of Shallow Underground Sources Based On Inverse Time Energy Focusing

Shallow underground microseismic positioning is the core technology in the field of target positioning in underground space.It is a civilian demand for coal mine survey,geological monitoring,engineering blasting,anti-theft monitoring of cultural relics key technology.At present,the location of shallow underground sources is mainly based on deep natural seismic positioning methods,but compared with the large-area,large-equivalent,large-depth,and long-term source positioning,the shallow underground source positioning has the following characteristics: 1)The depth of the underground source is shallow,generally No more than 100 m,the shallow geological structure is complex and unknown,and it is difficult to characterize the formation velocity parameters;2)Various shear waves and longitudinal waves in the shallow underground seismic field overlap each other,and the near-field dispersion phenomenon of the source is obviously aliased seriously,and the positioning feature parameters are extracted The difficulty is high,resulting in low accuracy of energy focus point reconstruction and difficulty of focus point recognition,which ultimately leads to low accuracy of source positioning.In order to solve the above problems,this paper mainly constructs the underground three-dimensional energy field by the counter-time amplitude superposition method,and combines the advantages of deep learning to achieve high-precision source target positioning in the underground shallow space.First,construct the underground three-dimensional energy field by amplitude superposition method,and by analyzing various imaging mechanisms,in the case of a small number of sensor nodes,improve the resolution of the energy field by grouping cross-correlation imaging method;secondly,the The three-dimensional energy field sequence is input to the 3D-CNN convolutional neural network for training and testing to improve the generalization of the network model;finally,a comparative analysis with the traditional simulated annealing algorithm verifies that the deep learning method proposed in this paper is superior to simulation Annealing algorithm.The research in this paper has been verified by field test,and the positioning error is 0.15m-0.20 m within the range of 100*100*50m.When using the deep learning network model for testing,the model accuracy is stable to 87%.The experimental results show that the energy focusing method proposed in this paper can be applied to typical shallow underground geological structures.The shallow energy model solution method based on deep learning model can effectively improve the location accuracy of shallow underground seismic sources.