基于PSA引导双分支神经网络特征融合的同步电机故障诊断  

Fault Diagnosis of Synchronous Motor Based on PSA Guided Double Branch Neural Network Feature Fusion

作　　者：李俊卿[1] 苑浩 黄涛 张承志 何玉灵[3] 张波[1] LI Junqing;YUAN Hao;HUANG Tao;ZHANG Chengzhi;HE Yuling;ZHANG Bo(Department of Electric Power Engineering,North China Electric Power University,Baoding071003,China;State Grid Puyang Power Supply Company,Puyang 457000,China;Department of Mechanical Engineering,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学电力工程系,河北保定071003 [2]国网濮阳供电公司,河南濮阳457000 [3]华北电力大学机械工程系,河北保定071003

出　　处：《智慧电力》2024年第12期51-58,共8页Smart Power

基　　金：国家自然科学基金资助项目(52177042)。

摘　　要：针对单一传感器信号在同步电机故障诊断中精度不高的问题,提出了1种基于金字塔切分注意力机制(PSA)的神经网络模型。首先,将三相电流信号和振动信号作为双分支输入到卷积神经网络进行特征提取,之后通过特征融合层将提取的信号特征进行融合。其次,添加PSA注意力机制捕获不同尺度的空间信息来丰富特征空间。最后,通过输出层输出诊断结果。实验表明所提模型能够显著提升同步电机故障诊断的准确率。Targeting the problem of poor accuracy of single sensor signal in synchronous motor fault diagnosis,the paper proposes a pyramid split attention(PSA)based neural network model.Firstly,the three-phase current signal and vibration signal are input into the convolutional neural network as two branches for feature extraction,and the extracted signal features are fused through the feature fusion layer.Secondly,the spatial information of different scales is captured with PSA attention mechanism to enrich the feature space.Finally,the diagnosis results are output through the output layer.The experiments show that the proposed model can significantly improve the accuracy of the synchronous motor fault diagnosis.

关 键 词：同步电机 PSA注意力机制 双分支特征融合 故障诊断 神经网络 

分 类 号：TM341[电气工程—电机]