| 引用本文: | 侯鹏飞,马宏忠,吴金利,张俊杰.基于混沌理论与蝗虫优化K-means聚类算法的电抗器铁芯和绕组松动状态监测[J].电力自动化设备,2020,40(11): |
| HOU Pengfei,MA Hongzhong,WU Jinli,ZHANG Junjie.Looseness status monitoring of reactor core and winding based on chaos theory and K-means clustering algorithm optimized by grasshopper algorithm[J].Electric Power Automation Equipment,2020,40(11): |
|
| 摘要: |
| 为了更加准确有效地监测高压并联电抗器铁芯和绕组机械状态，提出了基于混沌理论与蝗虫优化K-means聚类算法的电抗器铁芯和绕组松动状态监测方法。首先，对振动信号的混沌特性进行分析，采用C-C法选择最佳延迟时间和嵌入维数，对电抗器振动信号进行相空间重构；然后，利用蝗虫算法优化传统K-means聚类算法，从而更加合理地选取初始簇中心，进而通过优化后的K-means聚类算法求出重构信号相轨迹的簇中心；最后，根据簇中心位移矢量和的模值变化对电抗器铁芯和绕组松动状态进行监测。研究结果表明：采用Wolf法求得的各测点最大Lyapunov指数均大于0，证明电抗器振动信号具有混沌特性。蝗虫优化K-means聚类算法有效提高了计算结果的准确性，振动信号相轨迹的簇中心位移矢量和的模值变化能够有效反映铁芯和绕组松动故障隐患，从而为电抗器铁芯和绕组松动状态检修提供了理论依据。 |
| 关键词: 高压并联电抗器 铁芯和绕组 蝗虫优化K-means聚类算法 混沌理论 振动信号 监测 |
| DOI：10.16081/j.epae.202009033 |
| 分类号:TM47 |
| 基金项目:国网江苏省电力有限公司重点科技项目(J2018014) |
|
| Looseness status monitoring of reactor core and winding based on chaos theory and K-means clustering algorithm optimized by grasshopper algorithm |
|
HOU Pengfei1, MA Hongzhong1, WU Jinli1, ZHANG Junjie2
|
|
1.College of Energy and Electrical Engineering, Hohai University, Nanjing 211100, China;2.Institute of Electrotechnology, Baoding Tianwei Baobian Electric Co.,Ltd.,Baoding 071056, China
|
| Abstract: |
| In order to monitor the core and winding mechanical status of high voltage shunt reactor more accurately and effectively, a monitoring method based on chaos theory and K-means clustering optimized by grasshopper algorithm is proposed. Firstly, the chaotic characteristics of the vibration signal are analyzed, and the optimal delay time and embedding dimension are selected by C-C method to fulfil the phase space reconstruction of reactor vibration signal. Then the grasshopper algorithm is used to optimize the traditional K-means clustering algorithm, so as to select the initial cluster center more reasonably, and the optimized K-means algorithm is used to obtain the cluster center of reconstructed signal phase trajectory. Finally, according to the mode value change of displacement vector sum of cluster centers, the looseness status of reactor core and winding is monitored. The results show that the maximum Lyapunov exponents of each measuring point obtained by Wolf method are greater than zero, which proves that the reactor vibration signal has chaotic characteristics. The K-means clustering algorithm optimized by grasshopper algorithm effectively improves the accuracy of the calculation results. The mode value change of displacement vector sum of cluster centers of the vibration signal phase trajectory can effectively reflect the hidden danger of core and winding looseness, thus providing a theoretical basis for the looseness state maintenance of the reactor core and winding. |
| Key words: high voltage shunt reactor core and winding K-means clustering algorithm optimized by grasshopper algorithm chaos theory vibration signal monitoring |
