Summary
In Sweden, surface partial discharges (PD) in 24/0.4 kV secondary substations have been identified as a reliability concern, with the potential to cause equipment failures. This study used inspection data from over 500 substations, combined with asset and environmental information, to develop features for predicting PD-related faults in cable terminations.
Read more Read lessSeveral Machine learning (ML) algorithms were developed and trained to predict the likelihood of surface PD, enabling Distribution System Operators (DSOs) to target at-risk installations for preventive inspections. The models demonstrated high sensitivity and accuracy, successfully identifying all defective cable terminations in validation tests, with only a small number of false positives. This approach allows DSOs to optimise maintenance resources, prioritise interventions, and reduce the risk of costly failures.
The results show that data analytics and ML can significantly enhance asset management in power distribution networks, supporting more effective and proactive maintenance strategies.
Additional informations
| Publication type | Session Materials |
|---|---|
| Reference | B3_11425_2026 |
| Publication year | |
| Publisher | CIGRE |
| Country | Sweden |
| Study committees | |
| File size | 514 KB |
| Price for non member | 30 € |
| Price for member | 30 € |
Authors
LINDQUST Tommie - RISE; SANDELS Claes - RISE; JOHANSSON Tomas - Svensk Linjebesiktning
