| 引用本文: | 马宏忠,侯鹏飞,严岩,吴金利,陈轩,朱超.高压电抗器绕组和铁芯机械故障的混沌特性分析与特征识别[J].电力自动化设备,2022,42(5): |
| MA Hongzhong,HOU Pengfei,YAN Yan,WU Jinli,CHEN Xuan,ZHU Chao.Chaotic characteristic analysis and feature recognition of mechanical failure of high voltage reactor winding and iron core[J].Electric Power Automation Equipment,2022,42(5): |
|
| 摘要: |
| 首先，根据Kolmogorov熵判断高压并联电抗器振动信号的混沌特性，在证明其具有混沌特性的基础上重构振动信号，分析不同电压等级、不同故障类型和故障程度下振动信号相轨迹的变化规律。然后，将最大Lyapunov指数、Kolmogorov熵、关联维数作为一组混沌指标定量计算和表征高压并联电抗器绕组和铁芯发生机械故障前、后振动信号混沌特性。最后，利用多混沌指标构建特征空间进行故障识别，并与已有的高压并联电抗器故障识别方法进行对比验证多混沌特征空间识别故障的优越性。分析结果表明，多混沌特征空间能够准确实现无监督在线故障识别，其准确率比已有方法至少高5 %。 |
| 关键词: 电抗器 绕组 铁芯 故障识别 振动信号 混沌分析 特征空间 |
| DOI：10.16081/j.epae.202202012 |
| 分类号:TM47 |
| 基金项目:中央高校基本科研业务费专项资金资助项目(B200203128)；国网江苏省电力有限公司重点科技项目(J2018014)；江苏省研究生科研与实践创新计划项目(KYCX20_0431) |
|
| Chaotic characteristic analysis and feature recognition of mechanical failure of high voltage reactor winding and iron core |
|
MA Hongzhong1, HOU Pengfei1, YAN Yan1, WU Jinli1, CHEN Xuan2, ZHU Chao2
|
|
1.College of Energy and Electrical Engineering, Hohai University, Nanjing 211100, China;2.Maintenance Branch Company of State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing 211102, China
|
| Abstract: |
| Firstly, the chaotic characteristics of the vibration signals of high-voltage shunt reactor are judged according to the Kolmogorov entropy. On the basis of proving the chaotic characteristics of the vibration signals, the vibration signals are reconstructed, and the change law of phase trajectory under different voltage levels, fault types and fault levels is analyzed. Then, the maximum Lyapunov exponent, Kolmogorov entropy and correlation dimension are used as a set of chaotic indices to quantitatively calculate and characterize the chaotic characteristics of the vibration signals before and after the mechanical fault of high-voltage shunt reactor’s winding and iron core. Finally, the multiple chaotic indices are used to construct a feature space for fault identification, and the superiority of multiple chaotic feature space in fault identification is verified by comparing with existing fault identification methods. The verification results show that the multi-chaotic feature recognition space can accurately realize unsupervised online fault recognition, and its accuracy rate is at least 5 % higher than that of the existing methods. |
| Key words: electric reactors electric windings iron cores fault identification vibration signal chaos analysis feature space |
