Failure probability metric by machine learning for online risk assessment in distribution networks

Abstract

The risk assessment approach is useful for monitoring and supervisory control because it provides distribution operator with the capability to quantify the tradeoff between reliability and economic performance. The risk assessment determines the likelihood of something going wrong in a distribution network through the failure probability metric. To deal with the massive variety of information required in the calculation of failure probability we propose a data mining approach. The proposed approach incorporates weather, asset and outage information for characterizing the risk in a distribution network section via GIS platform.