基于数字孪生的复杂矿用设备预测性维护系统

针对恶劣工况环境下复杂矿用设备状态监测与预测性维护困难等问题，结合状态监测、故障预警和预测维护等多种综合建模与分析预测技术，提出了一种基于数字孪生的预测性维护系统。首先，介绍了复杂矿用设备数字孪生体的设计流程与构建原理，并在搭建数字孪生体的过程中实现了预测性维护系统的功能；然后，研究了基于LabVIEW、MySQL和Unity3D的状态数据获取方法，利用Unity3D开发引擎搭建了三维可视化复杂矿用设备状态监测平台，并通过虚拟空间可视化展示设备当前状态；最后，分析了优化BP（back propagation，反向传播）神经网络在复杂矿用设备故障预警中的适用性，同时利用MATLAB软件建立了复杂矿用设备关键零部件的预测性维护模型，并将预警结果通过MySQL数据库传输至Unity3D开发引擎，以驱动和部署预设维护流程，实现设备状态实时监测下的关键零部件故障预警。根据采煤机液压系统的实际维修流程，制定了混合现实（mixed reality, MR）预测性维护策略，并以采煤机摇臂部液压柱塞泵为实验对象开展有效性验证。结果表明，所构建的预测性维护系统的故障预测准确率高于90%，且故障预警结果可驱动HoloLens眼镜实现虚拟指导的维修交互，验证了该系统预测性维护功能的有效性。研究结果可为复杂矿用设备的预测性维护提供新思路。

Predictive maintenance system for complex mining equipment based on digital twin

Aiming at the difficulties of status monitoring and predictive maintenance of complex mining equipment under harsh working conditions, a predictive maintenance system based on digital twin was proposed by combining various comprehensive modeling, analysis and prediction techniques, such as status monitoring, fault warning and predictive maintenance. Firstly, the design flow and construction principle of digital twin of complex mining equipment were introduced, and the function of predictive maintenance system was realized in the process of building digital twin. Then, the status data acquisition method based on the LabVIEW, MySQL and Unity3D was studied, and the Unity3D development engine was used to build a three-dimensional visual status monitoring platform for complex mining equipment, and the current equipment status was visualized through virtual space. Finally, the applicability of optimized BP (back propagation) neural network in the fault warning of complex mining equipment was analyzed. At the same time, the predictive maintenance model of key parts of complex mining equipment was established by MATLAB software, and the warning results were transmitted to the Unity3D development engine through MySQL database to drive and deploy the preset maintenance process, so as to achieve the fault warning of key parts under the real-time monitoring of equipment status. According to the actual maintenance process of shearer hydraulic system, a mixed reality (MR) predictive maintenance strategy was formulated, and the effectiveness of the hydraulic plunger pump of shearer rocker arm was verified by taking it as the experimental object. The results showed that the fault prediction accuracy of the proposed predictive maintenance system was higher than 90%, and the fault warning results could drive HoloLens glasses to achieve the maintenance interaction of virtual guidance, which verified the effectiveness of the predictive maintenance function of the system. The research results can provide new ideas for predictive maintenance of complex mining equipment.