摘要
针对金属矿山破碎机故障诊断中传统方法误报率高的问题,研究提出基于卷积神经网络(CNN)的智能诊断模型。实验结果表明,CNN对轴承失效、衬板磨损/脱落、转子不平衡及传动系统故障的识别准确率分别达95.0%、93.3%、96.7%和95.0%,整体准确率95.0%,较SVM(90.8%)和决策树(89.2%)显著提升。该模型通过端到端特征学习机制,有效捕捉非线性耦合故障特征,为矿山设备预测性维护提供了高精度解决方案。
Addressing the high false alarm rate of traditional methods in fault diagnosis for crushers in metal mines,this study proposes an intelligent diagnostic model based on convolutional neural networks(CNN).Experimental results demonstrate that CNN achieves recognition accuracies of 95.0%,93.3%,96.7%,and 95.0% for bearing failure,liner wear/detachment,rotor imbalance,and transmission system faults respectively,with an overall accuracy of 95.0%.This represents a significant improvement over Support Vector Machines(SVM,90.8%)and decision trees(89.2%).Through its end-to-end feature learning mechanism,this model effectively captures nonlinear coupled fault characteristics,providing a high-precision solution for predictive maintenance of mining equipment.
作者
李群
Li Qun(Chifeng Jinxin Mining Co.,Ltd.,Sinosteel Group,Chifeng Inner Mongolia 024022,China)
出处
《机械管理开发》
2025年第12期87-88,91,共3页
Mechanical Management and Development
关键词
矿山机械
故障诊断
机器学习
mining machinery
fault diagnosis
machine learning
