摘要
针对异步电机在强干扰下故障诊断的挑战,多传感器融合虽能提升精度,但异构传感器数据的有效融合及传感器间相关性与物理先验知识的利用仍是难题。提出一种基于图神经网络(GNN)的异构多传感器融合交流电机故障诊断方法。采用多任务增强自编码器提取节点特征以学习判别性表示;构建融合物理先验约束的邻接矩阵以增强泛化性;利用图同构网络(GIN)获取图级表示进行分类。实验结果表明,该模型在两个异构异步电机数据集上的故障诊断性能优于传统方法。结论证实,所提框架通过有效的数据融合与物理知识整合,显著提升了故障诊断的准确性与鲁棒性。
To address the challenge of fault diagnosis for asynchronous motors under strong interference,multi-sensor fusion can improve accuracy,but effective fusion of heterogeneous sensor data and utilization of inter-sensor correlations and physical prior knowledge remain difficult problems.An AC motor fault diagnosis method based on heterogeneous multi-sensor fusion with graph neural network(GNN)is proposed.A multi-task enhanced autoencoder is employed to extract node features to learn discriminative representations;an adjacency matrix incorporating physical prior constraints is constructed to enhance generalization;and a graph isomorphism network(GIN)is used to obtain graph-level representations for classification.Experimental results show that the proposed model outperforms traditional methods in fault diagnosis performance on two heterogeneous asynchronous motor datasets.The conclusion confirms that the proposed framework significantly improves the accuracy and robustness of fault diagnosis through effective data fusion and integration of physical knowledge.
作者
郭辉
陈友兴
GUO Hui;CHEN Youxing(School of Intelligent Manufacturing and Vehicles,Shanxi Jinzhong Institute of Technology,Jinzhong 030600,China;School of Information and Communication Engineering,North University of China,Taiyuan 030051,China)
出处
《传感器与微系统》
北大核心
2026年第4期96-100,共5页
Transducer and Microsystem Technologies
基金
山西省回国留学人员科研资助项目(省级重点项目,2022-145)。
