摘要
“三下”开采引发的沉陷变形对输电线路等地表建(构)筑物的安全构成威胁,亟需一种采空区沉陷变形早期感知与智能预警方法。为此,提出了一种基于微震事件频次的采空区沉陷变形智能预警框架。该框架利用分布式声波传感(distributed acoustic sensing,简称DAS)系统采集微震数据,通过STA/LTA算法提取微震事件,借助结合自编码器(autoEncoder,简称AE)与高斯混合模型(Gaussian mixture models,简称GMM)的深度聚类方法对其分类,基于微震事件频次与沉陷变形数据之间的相关系数筛选诱发沉陷变形的微震事件,进而采用VGG-16深度学习模型实现对这类微震事件的智能识别,通过设定预警阈值开展实时预警。以我国西部某典型煤矿采空区为例,将该框架应用于实地监测。结果表明,该框架将采集到的采空区微震事件分为五大类,从中提取出诱发沉陷变形的一类微震事件,结合微震事件智能识别模型,成功对沉陷变形引发的杆塔倾斜度激增事件作出预警。实例证明该框架能够有效捕捉微震事件与沉陷变形的关联,实现对采空区沉陷变形的预警,具有实践可行性和工程应用价值。
[Objective]The subsidence deformation caused by the"three underground"mining pose a threat to the safety of surface buildings such as transmission lines,and there is an urgent need for an early perception and intelligent warning method for subsidence and deformation in goaf areas.[Methods]This paper proposed an intelligent early warning framework for subsidence and deformation in goaf areas based on the frequency of microseismic events.This framework utilized a Distributed Acoustic Sensing(DAS)system to collect microseismic data,extracted microseismic events using the STA/LTA algorithm,and classified the microseismic events using a deep clustering method that combined AutoEncoder(AE)and Gaussian Mixture Models(GMM).Based on the correlation coefficient between microseismic event frequency and subsidence deformation data,microseismic events that induced subsidence deformation were selected.The VGG-16 deep learning model was then used to achieve intelligent recognition of such microseismic events,and real-time warning was carried out by setting warning thresholds.[Results]This paper took a typical coal mine goaf in western China as the research area and applied the framework to field monitoring.The results show that the framework classifies the collected microseismic events in goaf into five categories,extracts one type of microseismic event that induces subsidence deformation,and combines with an intelligent microseismic event recognition model to successfully issue a warning for the sudden increase in tower inclination caused by subsidence deformation.[Conclusion]Therefore,this framework can effectively capture the correlation between microseismic events and subsidence and deformation,to achieve early warning of subsidence and deformation in goaf areas,and has practical feasibility and engineering application value.
作者
曹凯
卢渊
庞小龙
贺志华
于晓清
王玄
CAO Kai;LU Yuan;PANG Xiaolong;HE Zhihua;YU Xiaoqing;WANG Xuan(EHV Company of State Grid Ningxia Electric Prower Co.,Ltd.,Yinchuan 750011,China;Power Transformer Engineering Research Institute,China Electric Power Research Institute,Beijing 102401,China;State Grid Electric Prower Engineering Institute Co.,Ltd.,Beijing 100069,China)
出处
《地质科技通报》
北大核心
2025年第4期78-89,共12页
Bulletin of Geological Science and Technology
基金
国网宁夏电力有限公司科技项目“采动影响区重要输电线路杆塔周边地质监测预警与基础变形弹性防治快速矫正技术研究”(5229CG230008)。
关键词
煤层采空区
沉陷变形
分布式声波传感
微震信号
深度学习
智能预警
coal seam goaf
subsidence deformation
distributed acoustic sensing
microseismic signal
deep learning
intelligent early warning
