针对淮河能源集团当前开采深部A组煤，受底板水害严重威胁而缺乏有效的智能化、全覆盖预警技术的问题，提出了构建基于水文、充水水源及底板破坏实时监测物联网的多源信息大数据智能预警云平台而指导矿区内突水灾害事故预防的技术思路。以张集煤矿回采A组煤的1612A工作面为工程背景，构建了观测水源、水位、水压、水质、水温等参数变化的水文监测物联网数据传输系统，实时数据采集、数据传输、数据分析处理。而针对底板破坏深度则构建了微震实时监测物联网，传感器拾取底板破裂信号通过井上、井下的光纤环网传输到地面数据信号处理终端，实时监测反演底板破坏、导水裂隙通道分布。将回采、地质和监测数据各类因素考虑在内，建立了基于神经网络和深度学习的预警模型，确定了煤矿安全评价的主控指标以及评判指标，搜集全国大量矿井数据对模型进行了机器学习训练。集成数据采集、管理配置、设备监测、中控大屏、多维分析和故障预警5个模块，由网络集成技术和数据整合技术实现无缝连接，建立了煤矿底板突水灾害大数据预警云平台系统。将学习后的预警模型嵌入系统，基于整合的多源数据进行底板突水危险性评估与预警，与微震数据互馈分析实时发布决策信息。最后，选定淮南等矿区内的5组矿井数据进行了评估验证，并对1612A工作面回采进行了预警分析。结果表明，选定的矿井计算预测结果合理；张集矿工作面发生突水灾害的概率较小，监测期内未发生预警信息；证实了大数据云平台在煤矿底板水害防治应用的可行性，为淮河能源集团开采A组煤预防底板水害提供了新的技术支撑。

In view of the problems of effective intelligent and full coverage early warning technology，and the serious threat from floor water disaster when mining the deep coal from Gourp A in Huaihe Energy Group，the technical idea of guiding the prevention and control of water inrush in mining area was proposed.The big data intelligent early-warning cloud platform was built from the multi-source information based on the real-time monitoring of hydrology，water source，and floor failure.Based on the engineering background of the 1612A working face of Group A in Zhangji coal mine，the internet of things of hydrological monitoring data transmission system was constructed to observe the changes of water source，water level，water pressure，water quality，water temperature，and other parameters，which was used for real-time data collection，data transmission，data analysis，and processing.The micro-seismic monitoring internet of things was built for estimating failure depth of the floor.The sensor can pick up the floor failure signal and transmit it to the ground data signal processing terminal through the optical fiber ring network of the mine.The real-time monitoring and inversion of the floor failure，and the distribution of the water conducting fracture channel can be carried out.Taking various factors of mining，geology，and monitoring data into account，an early-warning model based on neural network and in-depth learning was established.The main control index and evaluation index of coal mine safety evaluation were determined，and machine learning training was carried out for the model by collecting a large amount mine data nationwide.A big data early-warning cloud platform system for water inrush disaster from coal mine floor was established，and the system integrates five modules:data collection，management configuration，equipment monitoring，central control panel，multi-dimensional analysis，and fault early warning.The seamless connection was realized by network integration and data integration technologies.The learned early-warning model was embedded in the system.Based on the integrated multi-source data，the risk assessment and early warning of water inrush from the mine floor were carried out，and the decision-making information was released in real time through mutual feed analysis with micro-seismic data.At last，five groups of data of coal mines in Huainan and other mining areas were selected to verify the model.The evaluation and analysis were carried out in the 1612A working face.The results show that the calculation and prediction results of chosen mines were reasonable，and the probability of water inrush disaster in the Zhangji coal mine was small.There was no early warning information in the monitoring period，which confirmed the feasibility of the application of big data cloud platform in the prevention and control of coal mine floor water disaster.It provides a new technical support for monitoring and early warning of mine floor water disaster in mining deep coal in Huaihe Energy Group.

1 工程概况

2 底板水害监测系统建设

   2.1 水文监测系统

   2.2 微震监测系统

3 大数据预测预警平台

   3.1 基于神经网络的预警模型

   3.2 预测预警平台构架

4 预测预警应用

   4.1 预警模型训练学习与验证

   4.2 预测预警

5 结论