Intermittent Earth Fault Detection In Distribution Network Based On The Voting Classification Technique

Abstract

Intermittent earth faults in non-effectively distribution networks, especially with underground cabling, can compromise the quality of the electricity supply. This type of earth fault may be followed by permanent faults, which in turn puts the networks on the line. This phenomenon monitoring can help distribution system operators (DSOs) to plan maintenance to reduce system interruption and improve MV electricity delivery. Thus, this research will examine AI-driven approaches, which are suitable for complicated issues, to improve distribution power grid monitoring and maintenance. The research focuses on medium and low-voltage grids and applies the voting classification technique (VC) to monitor and predict earth faults. Moreover, IEC 61850 Sampled Value communication protocol will be utilized at a practical level to establish a hierarchical infrastructure of data processing nodes. VC will process this raw data to determine the distribution network condition. In this endeavor, a new efficient way to monitor and maintain power networks will be examined. The suggested method will predict the existing and future status of the system, including upcoming breakdowns. At the top of the structure, aggregated information will be displayed to human grid operators to help them schedule maintenance or plan emergency actions. Real grid pilots and laboratory experiments in Finland will provide the required data to develop and train the suggested approach to predict intermittent earth faults.