Event Zone Identification in Electric Utility Systems Using Statistical Machine Learning

The next level of waveform analytics in power and energy systems involves sensitive event capturing and processing where some of the attributes such as event zone are not known apriori. A decision-tree-based methodology for identifying the origin of a general abnormality is discussed in this paper with deployment results from a multi-feeder distribution system. Field validation results highlight the opportunities for enhanced outage and asset management as well as practical challenges associated with applying this level of analytics in resource-constrained edge computing environments such as field controllers and smart sensors.