PD pattern recognition for GIS based on CNN architecture optimized by the double attention mechanism
Author:
Affiliation:

Clc Number:

TM85

Fund Project:

  • Article
  • |
  • Figures
  • |
  • Metrics
  • |
  • Reference
  • |
  • Related
  • |
  • Cited by
  • |
  • Materials
  • |
  • Comments
    Abstract:

    The PD pattern recognition is an important part of the insulation state evaluation of GIS. For the purpose of the accurate and efficient identification of PD signals, a new method of PD signal pattern recognition is proposed based on CNN optimized by the dual attention mechanism in this paper. Firstly, the GIS PD test platform is built, and four typical defects are set up in GIS chamber. The PD signals of different defects are collected by the UHF and ultrasonic detection respectively. Then, the data preprocessing is carried out based on the characteristics of the data obtained by methods mentioned above respectively. The feature space composed by the image features of UHF PD spectrum and the gram angle field density distribution of ultrasonic signal are constructed. Finally, the input image is extracted through the method of convolutional neural network optimized by double attention mechanism, and the results are predicted by a softmax classifier at the end of the network. It is shown that 97.57% recognition accuracy can be achieved by the fusion algorithm, which is higher than the recognition rate considering the single feature. The convolutional neural network optimized by the double attention mechanism is superior to the common algorithm in the aspects of the recognition rate, training speed and robustness.

    Reference
    Related
    Cited by
Get Citation

王彦彪,陈振勇,郭文萍,王宗宝,黄银汉.基于双注意力机制优化CNN架构的GIS局部放电模式识别[J].电力科学与技术学报英文版,2022,37(2):22-29. WANG Yanbiao, CHEN Zhenyong, GUO Wenping, WANG Zongbao, HUANG Yinhan. PD pattern recognition for GIS based on CNN architecture optimized by the double attention mechanism[J]. Journal of Electric Power Science and Technology,2022,37(2):22-29.

Copy
Share
Article Metrics
  • Abstract:
  • PDF:
  • HTML:
  • Cited by:
History
  • Received:
  • Revised:
  • Adopted:
  • Online: May 26,2022
  • Published: