高应力围岩灾变信息采集及智能预警  

Information Collection and Intelligent Warning for High Stress Surrounding Rock Disaster

作　　者：金声尧 贾澎涛[2] 王鹏 杨继红 侯长民 JIN Shengyao;JIA Pengtao;WANG Peng;YANG Jihong;HOU Changmin(Institute of Modern Coal Mining Technology,Shaanxi Coal Chemical Industry Technology Research Institute Company Limited,Xi’an,Shaanxi 710065,China;School of Computer Science and Technology,Xi’an University of Science and Technology,Xi’an,Shaanxi 710054,China;Shaanxi Coal Industry Company Limited,Xi'an,Shaanxi 710065,China;Xian Jumor Information and Technology Company Limited,Xi’an,Shaanxi 710076,China)

机构地区：[1]陕西煤业化工技术研究院有限责任公司现代煤炭开采技术研究所,陕西西安710065 [2]西安科技大学计算机科学与技术学院,陕西西安710054 [3]陕西煤业股份有限公司,陕西西安710065 [4]西安峻茂信息科技有限公司,陕西西安710076

出　　处：《矿业研究与开发》2021年第6期170-175,共6页Mining Research and Development

基　　金：国家重点研究发展计划项目(2018YFC0808303);国家自然科学基金项目(51974236);西安市科技计划项目(2020KJRC0069)。

摘　　要：针对煤矿围岩应力在线监测困难、灾害预警能力较弱的问题,通过自主设计的基于LoRa协议的围岩应力无线传感器和无线网关,实现了对井下围岩应力灾变的信息采集与数据传输。在围岩应力实时监测数据的基础上,应用深度学习理论,提出了应用自适应矩估计优化算法的深度循环神经网络围岩灾变预警模型(Adam-DRNN),构建了高应力围岩灾变信息捕捉及智能预警系统。试验结果表明,Adam-DRNN预警模型具有较高的准确度和较强的泛化能力,与BP神经网络预警模型、支持向量回归预警模型和差分自回归移动平均预警模型相比较,预测效果分别提高了23.1%、2%和6.1%。高应力围岩灾变信息捕捉及智能预警系统实现了对工作面及巷道区域矿山压力的实时监控和智能预警,提高了矿井安全生产管理水平,具有较强的实用价值。Aiming at the difficulties of online monitoring of surrounding rock stress and the weak disaster warning capabilities of coal mine, the self-designed wireless sensor and wireless gateway for surrounding rock stress based on the LoRa protocol were designed to realize the information collection and data transmission of underground surrounding rock stress disaster. On the basis of real-time monitoring data of surrounding rock stress, a deep recurrent neural network surrounding rock disaster warning model(Adam-DRNN) using adaptive moment estimation optimization algorithm was proposed by deep learning theory. Then, a system of information collection and intelligent warning for high stress surrounding rock disaster was constructed. The experimental results show that the Adam-DRNN warning model has high accuracy and strong generalization. Compared with the BP neural network warning model, the support vector regression warning model and the autoregressive integrated moving average warning model, its predictIve effect is increased by 23.1%, 2% and 6.1% respectively. This system realizes real-time monitoring and intelligent warning of mine pressure on working face and tunnel area, improves the level of mine safety production management, and has a good practical value.

关 键 词：围岩灾变 信息采集 无线传感器 循环神经网络 智能预警 

分 类 号：TD326[矿业工程—矿井建设]