Research of technology and system of tunnel microseismic monitoring and rockburst early warning based on deep learning
Relying on microseismic monitoring,deep learning and virtual simulation technology,a system and platform for the automatic integrated processing of tunnel microseismic information and intelligent warning of rock bursts is established in this paper.Both a microseismic multi-classification model based on bimodal feature extraction,and a dual-task model of noise reduction and arrival pickup of a waveform based on deep convolutional encoding and decoding network are proposed,and a microseismic positioning algorithm based on the gravity search method is put forward,for realizing automatic,efficient and accurate processing of tunnel microseismic classification,noise reduction,picking,positioning and source parameter calculation.Selecting cumulative apparent volume and energy index source parameters as key indicators,a parallel sequence prediction model for microseismic parameters and a prediction and warning model for rock burst incubation stage based on LSTM multi-variant network are established,which achieves early warning of the current future state and time evolution of rock bursts.Meanwhile,the integration and display of tunnel site geographic information,geological models,tunnel models and disaster(microseismic)information are achieved based on the three-dimensional visualization framework Cesium,forming a tunnel microseismic monitoring and rock burst warning system that integrates microseismic information collection module,microseismic information cloud processing module,and rock burst prediction and warning module.The system is applied to the rock burst disaster section of the Daxiagu Tunnel of Ehan Expressway,achieving automatic,efficient and accurate processing of massive microseismic data,and verifying the effectiveness of the automatic integrated processing of tunnel microseismic information and the intelligent warning technology system for rock bursts.
rock mechanicstunnelmicroseismic monitoringrockburst warningdeep learningvirtual simulation
NETL
NSTL
万方数据
