Wave analytical techniques-based fault initiation detection at middle voltage distribution feeders

Abstract

This paper presents studies, findings and results of the fault initiation detection approach, which is based on wave analytical techniques at middle voltage distribution feeders of ADM Distribution Company (DisCo), Turkey. Predictive methods are implemented to determine root causes of fault initiations on the feeders, which cannot be detected with today’s modern SCADA and relay protection systems that are not designed to prevent failure. In response, artificial intelligence-based machine-learning techniques, which accumulate experience in the current and voltage waveforms related to the failures, are addressed in the study. Thereby, distribution companies will be able to prevent failures that can result in a power outage on a large scale. The algorithms are tested on a feeder at ADM DisCo. The performance of the approach is discussed based on measurements. The root causes of fault initiations on the feeders, which cannot be detected with today’s modern SCADA and relay protection systems, are identified successfully.

Keywords: Wave analytics, artificial intelligence, machine learning, distribution;