Summary
The analysis of partial discharge (PD) current pulse waveforms provides valuable information for assessing the condition of high-voltage insulation systems, but their use in continuous monitoring is limited by the large data volumes generated by high-resolution measurements.
Read more Read lessThis work investigates AI-based compression methods for PD current pulse waveforms using autoencoder-based models. In addition to single-channel compressions, multichannel measurements and fixed-length sequences of PD events are considered to exploit inter-channel redundancies and temporal dependencies. The results demonstrate that very high compression rates can be achieved while preserving signal structures, with joint multichannel compression improving reconstruction quality and sequential compression retaining temporal information.
Beyond reconstruction, the compressed representations can be used directly for PD analysis, enabling the extraction of diagnostically relevant parameters and Phase-Resolved Partial
Discharge (PRPD) -like visualizations without full waveform reconstruction. Overall, the proposed framework significantly reduces storage and transmission requirements for PD monitoring applications.
Additional informations
| Publication type | Session Materials |
|---|---|
| Reference | D1_12396_2026 |
| Publication year | |
| Publisher | CIGRE |
| Country | Germany |
| Study committees | |
| File size | 943 KB |
| Price for non member | 30 € |
| Price for member | 30 € |
Authors
TRUE Pascal - Hochschule für Technik und Wirtschaft Berlin; GRÄF Thomas - Hochschule für Technik und Wirtschaft Berlin; MENGE Matthias - Hochschule für Technik und Wirtschaft Berlin; PLATH Ronald - Technische Universität Berlin
