Prediction of rock burst risk level based on combination of physical indexes and deep learning
In order to explore the early-warning problem of rock burst under the background of intelligent mining of coal mine,taking 21181 working face of Qianqiu Mine in Henan Province as the background,an early-warning method of rock burst combining the physical indexes and deep learning was proposed based on data analysis.By analyzing the characteristics of the maximum value,trend and absolute value of relative change rate of each physical index before the occurrence of the big energy event,the corresponding comprehensive physical characteristics were obtained.The spatial distribution characteristics of seismic source were analyzed,then the coordinate attention mechanism was proposed according to their characteristics,and the seismic source coordinates were weighted to obtain the seismic source characteristics.The comprehensive physical charac-teristics and seismic source characteristics were weighted by adding the channel attention mechanism.The whole connection layer was used for classification to achieve the purpose of risk grade prediction,and finally the model was applied to the actual project.The results show that the early-warning method of rock burst combining the physical indexes and deep learning can achieve a high accuracy,and the research results can provide some reference for practical engineering.
rock burstphysical indexdata analysisseismic source characteristicsdeep learningcoordinate attention mechanism
万方数据
维普
