Predicting Weather-related Power Outages in Distribution Grid

Abstract

Improvements in monitoring and data collection practices provide opportunities for more comprehensive modelling and managing grid operations. At the same time, advanced data analysis methods should be able to address service quality degradation due to outages, weather patterns and asset-related performance.In this paper, we apply Machine Learning and Computational Intelligence methods for the analysis of power distribution system data and constructing a system for predicting power outages. Weather and outage data are utilized by the proposed system for predicting purposes. We evaluate the prediction performance of different types of prediction models. We also propose and validate three different architectures of a system for predicting types of weather-related outages. We focus on outages caused by wind, snow and ice. An analysis of the prediction results is provided.