This conductor can remain energised as it is effectively back fed through the magnetising impedance of downstream transformers, with insufficient fault current to be detected by classical HV protection systems.
Voltage data collected at the low-voltage side of downstream transformers are analysed, and converted into phasors for further analysis. Machine learning techniques, involving clustering of the data and identification of key features, is then used to develop a simple predictive analytics model that can be used to give a clear indication of such a ConductorDown event.
The predictive model is applied right at the edge on ElectroNet’s PowerPilot platform and this allows for maximum selectivity and sensitivity, while minimising the response time. An alarm is transmitted using the standard PowerPilot messaging protocol, and this can either alert control room staff so they can react to and isolate the affected section of line, or can trigger an automated response through an ADMS or SCADA system.