Fast Traveling Wave Detection and Identification Method for Power Distribution Systems Using the Discrete Wavelet Transform.

This work proposes a Traveling Wave (TW) detection and identification method that addresses the demanding time and functional constraints that TW-based protection schemes for power distribution systems require. The high-frequency components of continuously sampled voltage signals are extracted using the Discrete Wavelet Transform, and the designed indicator is monitored to detect the TW arrival time. The limitations of the method are explored, such as the effective range of detection and the exposure to TWs originating from non-fault events. Simulations are conducted on the IEEE 34 nodes system, which has been adapted to include capacitor banks and small loads connection events, as well as transformer energization and de-energization events. After the TW detection, a Random Forest classifier has been trained to infer whether the TW is due to a fault or another type of transient. About the results, the proposed method is sensitive to near faults, and faults can be successfully distinguished from other events.