Microseismic multi-parameter based rockburst early warning using dynamic Bayesian networks - RASIM 2022

As an advanced method of rock engineering construction safety monitoring, microseismic monitoring technology can monitor the micro-fracture of rock masses within an effective range, and then perform forecast analysis of rockburst, cracking, falling blocks and other disasters. Although the application of microseismic monitoring technology has become more mature, how to improve its application efficiency remains to be further studied. At present, artificial intelligence has been extensively developed, and the use of intelligent algorithms to improve the efficiency of rockburst forecast can provide a valuable reference for the safe construction of underground engineering. To this end, this paper proposes a microseismic multi-parameter rockburst early warning model based on dynamic Bayesian networks. First, the sensitivity of microseismic source parameters to rockburst forecast is analyzed based on microseismic monitoring data and actual rockburst information of a hydropower station. Second, the main seismic source parameters and their data are selected and imported into a dynamic Bayesian network for learning to obtain the rockburst forecast model. The model is confirmed in its self-validation and K times cross-validation, and the receiver operating characteristic curve of the model is also analyzed, it is found that the area under the curve is close to 1, indicating that the model has good accuracy. Finally, the analysis of the influence strength of the parent and child nodes in the model reveals that the moment magnitude and source scale are the most sensitive parameters, which can be used as an important reference for rockburst forecast.