基于卷积神经网络与SLAM的露天煤矿无人驾驶车辆故障诊断  

Fault Diagnosis of Driverless Vehicle in Open-pit Coal Mine Based on Convolutional Neural Networks and SLAM

作　　者：舒应秋 戚红建 李伟 孙涛 Shu Yingqiu;Qi Hongjian;Li Wei;Sun Tao(Huaneng Yimin Coal and Electricity Co.,Ltd.,Hulunbuir 021100,China;Huaneng Information Technology Co.,Ltd.,Qingdao 266000,China)

机构地区：[1]华能伊敏煤电有限责任公司,内蒙古呼伦贝尔021100 [2]华能信息技术有限公司,山东青岛266000

出　　处：《煤矿机械》2025年第5期175-178,共4页Coal Mine Machinery

摘　　要：针对露天煤矿无人驾驶车辆在复杂环境下故障诊断准确率低、实时性差的问题,提出一种融合卷积神经网络与同步定位与地图构建(SLAM)的故障诊断方法。首先,基于SLAM技术构建车辆运行环境的三维地图,实现车辆精确定位与环境感知,采集车辆状态、环境特征等多源数据;其次,设计多尺度卷积神经网络模型,对采集的故障特征数据进行深度学习,建立故障诊断模型;最后,在某露天煤矿无人驾驶矿车上进行实验验证。结果表明:该方法能够同时识别车辆系统、传感器、执行机构等多类故障,故障诊断准确率达到96.8%,平均诊断时间23.5 ms。In response to the problems of low accuracy and poor real-time performance in fault diagnosis of driverless vehicle in complex environments of open-pit coal mine,proposed a fault diagnosis method that integrates convolutional neural networks and simultaneous localization and mapping(SLAM).Firstly,based on SLAM technology,a three-dimensional map of the vehicle operating environment was constructed to achieve precise vehicle positioning and environmental perception,and collected multi-source data including vehicle status and environmental features.Secondly,designed a multi-scale convolutional neural network model to perform deep learning on the collected fault feature data and established a fault diagnosis model.Finally,experimental verification was conducted on driverless mining vehicle in a open-pit coal mine.The results show that this method can simultaneously identify multiple types of faults such as vehicle systems,sensors and actuators,with a fault diagnosis accuracy of 96.8%and an average diagnosis time of 23.5 ms.

关 键 词：露天煤矿 无人驾驶 故障诊断 卷积神经网络 SLAM 环境感知 

分 类 号：TD67[矿业工程—矿山机电]