悬臂式掘进机作为一种广泛使用的煤矿机械联合机组，具有截割、装载运输、自行走以及喷雾除尘功能。目前，煤矿掘进设备运行状态远程可视化监测大多通过视频进行人工监测，但井下掘进工作面粉尘大，环境复杂，状态多变，人工监测无法全面、准确、及时反映掘进机运行状态。当前煤矿开采及掘进装备正处于从“自动化+远程可视化干预”向“智能化+自适应控制”的关键过渡期，研制一种多源数据融合的掘进机运行状态智能监测系统，研究数字模型构建、数据感知与互联方法、基于BP神经网络的掘进机状态识别与预警方法等关键技术，提高悬臂式掘进机的监测水平，对构建智能掘进系统具有重要的研究意义和价值。

针对悬臂式掘进机人工监测效率低、可靠性差的问题，研究数据驱动的悬臂式掘进机虚拟仿真方法，通过构建悬臂式掘进机高精度数字模型，主要包含数字煤层模型、数字掘进机模型等，基于操控数据对模型进行驱动，在虚拟场景中对掘进动作进行还原分析，为掘进机状态监测提供直观显示。

针对悬臂式掘进机工作过程中截割载荷、截割头和机身位姿等数据采集困难、准确性差的问题，提出基于多传感器的掘进机截割载荷和位姿监测方法，研究了掘进机多源数据融合及高效实时的数据交互方法，保证了数据准确完整，为实现运行状态智能监测提供支持。

研究数字模型与数据融合的运行状态监测及异常预警可行性，提出基于BP神经网络的掘进机状态识别方法，构建基于BP神经网络的掘进机状态诊断与异常预警方法，将故障数据用于状态诊断算法训练，实现掘进机异常状态预警。

最后，搭建运行状态监测试验平台，对数字掘进仿真、状态识别及异常预警两个核心功能进行了应用验证，实验表明，平台能够在多源数据的驱动下实现数字掘进机与实体同步，可对掘进机故障进行诊断，对异常进行预警，符合设计需求。

As a widely used coal mine machinery combined unit, the cantilever roadheader has the functions of cutting, loading and transportation, self-walking and spray dust removal. At present, the remote visual monitoring of the running state of coal mine tunneling equipment is mostly monitored by video, but the underground tunneling face dust is large, the environment is complex and the state is changeable, so the manual monitoring cannot reflect comprehensively, accurately and timely the running state of the boring machine. Current coal mining and tunneling equipment is from "automation + remote visual intervention" to "intelligent + adaptive control" key transition, develop a multi-source data fusion machine running state intelligent monitoring system, the digital model construction, data perception and interconnection method, based on BP neural network machine fault diagnosis and early warning method and other key technology, improve the monitoring level of cantilever machine, to build intelligent tunneling system has important research significance and value.

Aiming at the problems of low efficiency and poor reliability of manual monitoring of Cantilever Roadheader, research data driven cantilever machine virtual simulation method, through, build cantilever machine high precision digital model, mainly contains digital coal seam model, digital boring machine model, based on the control data drive model, in the virtual scene, restore tunneling action analysis for boring machine state monitoring.

For cantilever roadheader work in the process of cutting load, cutting head and the fuselage position of data acquisition difficulty, poor accuracy, put forward based on multi-sensor roadheader cutting load and position monitoring method, studied the roadheader multi-source data fusion and efficient real-time data interaction method, ensure the data accurate and complete, to realize the running state intelligent monitoring.

 The feasibility of operation state monitoring and abnormal early warning of digital model and data fusion is studied, the fault diagnosis method of roadheader based on BP neural network is proposed, the condition diagnosis and abnormal early warning method based on BP neural network is constructed, and the fault data is used for state diagnosis algorithm training to realize the early warning of abnormal state of roadheader.

Finally, the operating condition monitoring test platform was built, and the two core functions of digital tunneling simulation, state recognition and anomaly early warning were applied to verify. The experiment showed that the platform could realize the synchronization of digital tunneling machine and entity under the drive of multi-source data, and could diagnose the fault of tunneling machine and give early warning of anomalies, which met the design requirements.