Williams, Iwan, Haddad, M.  ORCID: https://orcid.org/0000-0003-4153-6146, Albano, M. ORCID: https://orcid.org/0000-0002-5486-4299, Reid, A. ORCID: https://orcid.org/0000-0003-3058-9007 and Wilson, G.
2026.
Partial discharge localisation by artificial intelligence classification.
Presented at: 2026 IEEE 6th International Conference on Dielectrics (ICD),
Southampton, UK,
21 - 25 June 2026.
2026 IEEE 6th International Conference on Dielectrics (ICD).
IEEE,
pp. 382-385.
10.1109/icd64907.2026.11665451
|
Abstract
Partial discharge monitoring is a commonly used engineering tool for condition assessment of high voltage assets. In air insulated substations, radiofrequency antennas can be used to detect and localise partial discharge sources. For accurate partial discharge diagnosis, it is important that defect localisation is as accurate as possible. This paper proposes a neural network classification-based approach for localisation using radiofrequency measurements from multiple antennas. The proposed method relies on a ray tracing-based simulation of radiofrequency propagation to train the neural network. This approach is evaluated against a benchmark approach from a conventional received signal strength spatial localisation method. A hypothetical scenario with five possible partial discharge sources is utilised for the generation of test and evaluation data. An accuracy of 67% was achieved by the benchmark method, compared to 93% with the proposed method. This result indicates that the new approach could improve localisation of partial discharges in air insulated substations.
| Item Type: | Conference or Workshop Item - published (Paper) |
|---|---|
| Date Type: | Published Online |
| Status: | Published |
| Schools: | Schools > Engineering |
| Publisher: | IEEE |
| ISBN: | 9798331534929 |
| Last Modified: | 14 Sep 2026 16:00 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/189575 |
Actions (repository staff only)
 |
Edit Item |
Dimensions
Dimensions