关键词: 提升机, 故障诊断, 深度学习, 边缘节点, 诊断平台, 变工况

Abstract: Condition monitoring and on-line fault diagnosis of hoist in the actual industrial scene, insufficient labeled fault samples and the distribution difference between testing and training samples caused by variable working conditions are important reasons that restrict the application of intelligent fault diagnosis algorithm in practical engineering. This paper proposes a transfer fault diagnosis architecture combined edge node for actual hoist, including multi-source data perception layer, edge node layer, network layer and central cloud server layer. Taking the hoist bearing as the object, a deep transfer fault diagnosis algorithm based on ResNet and multi-core joint distribution discrepancy is proposed to realize the fault state recognition of hoist under variable working conditions, two kinds of bearing fault data are used to verify the effectiveness and adaptability of the algorithm. The results show that the proposed algorithm can achieve the ideal transfer fault diagnosis accuracy. Finally, a hoist detection and diagnosis platform is designed and constructed, which is deployed in the coal mine ground central cloud server center to realize the monitoring of the running state of the mine hoist and online diagnosis.