Storm Outage Prediction Using Machine Learning for Resiliency Enhancement

The increasing frequency and severity of extreme weather events, mostly as result of climate change, has accentuated the need for expanding traditionally defined restoration processes for utilities. Predicting the associated power outages before or during the firsts few hours of severe weather events can enable proactive preparation and response. In this paper, we develop an outage prediction model using machine learning algorithms. The outcome of the proposed model allows power system operators to make early efficient decisions about the preventive measures that will result in saving resources and minimizing costs to enhance the power system resilience. Employing real data from a utility company, the performance of the proposed model is examined and validated through numerical simulations resulting in a viable algorithm for predicting outages in presence of severe weather events.