振动条件下三相交流电动机线路故障电弧诊断方法  

Method for Diagnosing Line Fault Arcs in Three-Phase AC Motors Under Vibration Conditions

作　　者：孙益凡 刘玉军 张树旺 齐东迁 陈光华 郭凤仪 SUN Yifan;LIU Yujun;ZHANG Shuwang;QI Dongqian;CHEN Guanghua;GUO Fengyi(School of Electrical and Electronic Engineering,Wenzhou University,Wenzhou 325035,China;Kailuan(Group)Co.,Ltd.Qianjiaying Mining Branch,Tangshan 063300,China;Electro optic Explosion proof Technology Co.,Ltd.,Yueqing 325600,China)

机构地区：[1]温州大学电气与电子工程学院,浙江温州325035 [2]开滦(集团)有限公司钱家营矿业分公司,河北唐山063300 [3]电光防爆科技股份有限公司,浙江乐清325600

出　　处：《电器与能效管理技术》2024年第8期69-76,共8页Electrical & Energy Management Technology

基　　金：国家自然科学基金资助项目(52077158)。

摘　　要：串联故障电弧是由电气线路或设备中电气连接松动或线路绝缘老化引起的空气击穿现象,会导致设备损坏、电路短路等问题,甚至引起电气火灾。为了准确识别三相交流电动机的振动故障电弧,提出一个基于一维卷积神经网络(1D CNN)并结合鹈鹕优化算法(POA)的故障诊断模型。所提模型可以直接利用上位机输出的一维时序电流数据进行训练;通过POA寻找1D CNN的最佳超参数,提升模型的识别能力;经过t-SNE算法分析模型的特征提取效果后,证实其有效性。测试结果显示,故障电弧识别准确率高达99.72%,所提方法比现有技术更加方便且性能更优。Arc fault is an air breakdown phenomenon caused by loose electrical connections or aging insulation in wiring or equipment,which can lead to equipment damage,circuit faults,and even electrical fires.To accurately identify three-phase AC motor vibration fault arcs,a fault diagnosis model based on one-dimensional convolutional neural networks(1D CNN)combined with the pelican optimization algorithm(POA)is proposed.The proposed model can be trained directly using one-dimensional temporal current data output from the host computer.The POA is utilized to find the optimal hyperparameters of the 1D CNN,enhancing the model’s recognition capability.After analyzing the feature extraction effectiveness of the model using the t-SNE algorithm,its validity is confirmed.Test results show that the fault arc identification accuracy reaches 99.72%.The proposed method is more convenient and superior in performance compared to existing technologies.

关 键 词：卷积神经网络 三相交流故障电弧 鹈鹕优化算法 故障诊断 

分 类 号：TM34[电气工程—电机] TM501.2