Summary
Transmission line corridors traverse complex terrain and climatic conditions, making them prone to fault trips caused by natural factors. These faults severely threaten the safe and stable operation of the power grid as well as equipment reliability. Rapid, accurate, and reliable fault identification technology is crucial for improving line operation and maintenance. Distributed fault monitoring devices can capture traveling wave signals with distinct characteristics at the moment of a line fault, laying a foundation for accurate fault cause identification.
Read more Read lessThis study used 925 fault traveling wave data samples from 500kV and above transmission lines between 2018 and 2021, including 423 lightning faults and 502 non-lightning faults
(external damage, icing, windage yaw, bird damage, etc.). By deeply mining time-frequency waveform features, a deep learning algorithm integrating Convolutional Neural Networks
(CNN), Bidirectional Long Short-Term Memory networks (BiLSTM), and a Self-attention mechanism (CNN-BiLSTM-SelfAttention) is constructed for fault classification. After sample balancing, the overall fault recognition accuracy is improved from 82.6% to 96.2%. The proposed fast fault diagnosis method features high precision and reliability, which shortens fault recovery time, optimizes operation and maintenance, and supports transmission line accident prevention.
Additional informations
| Publication type | Session Materials |
|---|---|
| Reference | C4_11582_2026 |
| Publication year | |
| Publisher | CIGRE |
| Country | China, People's Republic of |
| Study committees |
|
| File size | 1 MB |
| Price for non member | 30 € |
| Price for member | 30 € |
Authors
XIE Yingpu - China Electric Power Research Institute; GU Shanqiang - China Electric Power Research Institute; ZHAO Chun - China Electric Power Research Institute; ZENG Yu - China Electric Power Research Institute; YAN Biwu - China Electric Power Research Institute; TAO Hantao - China Electric Power Research Institute
Keywords
