ABSTRACT The occurrence of mine quake is subject to coupling control of multiple factors. To study the influence of mine quake distribution on rock burst under different incentive conditions, based on massive microseismic monitoring data, this paper established time-varying signal analysis model for deep recurrent neural network and restricted Boltzmann machine deep process neural network, which supports multi-perspective and high-dimensional analysis of complex mine quake. For feature extraction of multi-source acoustic emission real-time monitoring signals and the discrimination of microquake and mine quake intensity, a deep neural network model was proposed for the identification of mine quake type and intensity. Practice has shown that the discrimination effect is fine.

中文翻译：

---

基于深度神经网络的矿山地震类型及强度判别分析

摘要 矿山地震的发生受多种因素的耦合控制。为研究不同激励条件下矿山地震分布对岩爆的影响，基于海量微震监测数据，建立了深度递归神经网络和受限玻尔兹曼机深度过程神经网络的时变信号分析模型，支持多复杂矿山地震的透视与高维分析[J]. 针对多源声发射实时监测信号的特征提取和微震和矿震强度的判别，提出了一种深度神经网络模型来识别矿震类型和强度。实践证明，区分效果良好。