Fault analysis based on high-resolution data acquisition is growing in use as it offers a more complete picture of faults, which provides an opportunity to deal with failures more effectively. However, with increasing volumes of data being collected, it becomes impossible for engineers to interpret every fault instance. To solve this, this thesis proposes novel power network fault detection and diagnosis methods applied to continuous high-frequency Power Quality (PQ) data. These novel methods deliver online anomaly segmentation, fault classification, and automatic fault labelling. The work addresses the need for increasing levels of situational awareness in distribution networks and its corresponding data-related challenges.The combination of these contributions can achieve automatic extraction of information from operational PQ data without excessive manual effort. This research uses simulated cases and operational data to validate the effectiveness of the contributions. The significance of this research is that it extracts critical information from continuous PQ data streams and automatically interprets the segmented signals, which reduces the demand for expert interpretation. In addition, it can operate through intensive monitoring at a single point on the network, which enhances the observability of the distribution network without installing excessive amounts of sensors.
|Date of Award||28 Jul 2020|
- University Of Strathclyde
|Sponsors||University of Strathclyde|
|Supervisor||Stephen McArthur (Supervisor) & Bruce Stephen (Supervisor)|