Graph Attention Network for Predicting Duration of Large-Scale Power Outages Induced by Natural Disasters

TL;DR

This paper introduces a Graph Attention Network-based model to predict power outage durations caused by natural disasters, effectively handling spatial dependencies and heterogeneity with high accuracy using limited data.

Contribution

The study presents a novel GAT-based approach with semi-supervised learning for outage duration prediction, outperforming existing models in real-world hurricane scenarios.

Findings

Achieved over 93% accuracy in outage duration prediction.

Outperformed XGBoost, Random Forest, GCN, and simple GAT by 2-15%.

Effective handling of spatial heterogeneity and data limitations.

Abstract

Natural disasters such as hurricanes, wildfires, and winter storms have induced large-scale power outages in the U.S., resulting in tremendous economic and societal impacts. Accurately predicting power outage recovery and impact is key to resilience of power grid. Recent advances in machine learning offer viable frameworks for estimating power outage duration from geospatial and weather data. However, three major challenges are inherent to the task in a real world setting: spatial dependency of the data, spatial heterogeneity of the impact, and moderate event data. We propose a novel approach to estimate the duration of severe weather-induced power outages through Graph Attention Networks (GAT). Our network uses a simple structure from unsupervised pre-training, followed by semi-supervised learning. We use field data from four major hurricanes affecting $501$ counties in eight Southeastern U.S. states. The model exhibits an excellent performance ($>93\%$ accuracy) and outperforms the existing methods XGBoost, Random Forest, GCN and simple GAT by $2\% - 15\%$ in both the overall performance and class-wise accuracy.