基于经验小波变换与相空间重构的故障电弧检测  

Fault arc detection based on empirical wavelet transform and phase space reconstruction

作　　者：刘天正 李春雨 王志明[1] 孙颀皓 卜雄洙[1] LIU Tianzheng;LI Chunyu;WANG Zhiming;SUN Qihao;BO Xiongzhu(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China;School of Electronic and Optical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学机械工程学院,南京210094 [2]南京理工大学电子工程与光电技术学院,南京210094

出　　处：《火灾科学(中英文)》2024年第4期263-270,共8页Fire Safety Science

基　　金：国家重点研发计划项目(2021YFC1523500)。

摘　　要：针对交流回路中串联故障电弧检测的困难问题,根据国标搭建试验平台采集四种典型负载的电流波形,使用经验小波变换分解提取特征分量。为了拓展单一特征分量中的故障电弧特征,引入相空间重构技术将一维特征分量嵌入到三维空间中,最后制成数据集输入到CNN-LSTM网络中进行训练,结果表明该方法可以准确地识别出故障电弧,识别准确率最高可达98.7%。结合通用工具型地理信息系统软件MHMapGIS,探析了将故障电弧识别技术传输到机器视觉监控系统,实现网格型可视化预警。Given the difficulty of detecting series arc faults in AC circuits,a test platform is set up according to the national standard to collect the current waveforms of four typical loads,and the feature components are extracted using empirical wavelet decomposition.Phase space reconstruction is introduced to embed the one-dimensional feature component into the three-dimensional space to expand the fault arc features in a single feature component.Finally,the data set is made into a CNN-LSTM network for training.The results show that the method can accurately recognize the fault arcs,and the highest recognition accuracy can reach up to 98.7%.Combined with the general tool-based geographic information system software MHMapGIS,this paper explores the transmission of fault arc recognition technology to machine vision monitoring systems,to achieve grid-based visual warning.

关 键 词：经验小波变换 维度重构 故障电弧 

分 类 号：TM501[电气工程—电器] X915.5[环境科学与工程—安全科学]