Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Predictive Modeling of Power Outages during Extreme Events: Integrating Weather and Socio-Economic Factors

Antar Kumar Biswas, Masoud H. Nazari

TL;DR

Predicts county-level, hourly outages during LPHC events by fusing EAGLE-I outage records with Open-Meteo weather, census socio-economic indicators, and infrastructure data. The framework uses SMOGN-based data balancing and trains RF, SVM, AdaBoost, and LSTM models, with LSTM delivering the lowest error on Michigan data ($MSE$ $=0.0086$, $MAE$ $=0.0432$) in the enhanced scenario that includes socio-economic and infrastructure features. Weather variables, especially precipitation and wind, are the primary drivers, while higher income and more developed infrastructure modestly reduce outage risk. The approach supports resilience planning and scenario analysis and can be extended to other regions and extreme-event types by incorporating localized data.

Abstract

This paper presents a novel learning-based framework for predicting power outages caused by extreme events. The proposed approach specifically targets low-probability, high-consequence outage scenarios and leverages a comprehensive set of features derived from publicly available data sources. We integrate EAGLE-I outage records (2014-2024) with weather, socio-economic, infrastructure, and seasonal event data. Incorporating social and demographic indicators reveals underlying patterns of community vulnerability and provides a clearer understanding of outage risk during extreme conditions. Four machine learning models (Random Forest (RF), Support Vector Machine (SVM), Adaptive Boosting (AdaBoost), and Long Short-Term Memory (LSTM)) are evaluated. Experimental validation is performed on a large-scale dataset covering counties in the lower peninsula of Michigan. Among all models tested, the LSTM network achieves the lowest prediction error. Additionally, the results demonstrate that stronger economic conditions and more developed infrastructure are associated with lower outage occurrence.

Predictive Modeling of Power Outages during Extreme Events: Integrating Weather and Socio-Economic Factors

TL;DR

Predicts county-level, hourly outages during LPHC events by fusing EAGLE-I outage records with Open-Meteo weather, census socio-economic indicators, and infrastructure data. The framework uses SMOGN-based data balancing and trains RF, SVM, AdaBoost, and LSTM models, with LSTM delivering the lowest error on Michigan data ( , ) in the enhanced scenario that includes socio-economic and infrastructure features. Weather variables, especially precipitation and wind, are the primary drivers, while higher income and more developed infrastructure modestly reduce outage risk. The approach supports resilience planning and scenario analysis and can be extended to other regions and extreme-event types by incorporating localized data.

Abstract

This paper presents a novel learning-based framework for predicting power outages caused by extreme events. The proposed approach specifically targets low-probability, high-consequence outage scenarios and leverages a comprehensive set of features derived from publicly available data sources. We integrate EAGLE-I outage records (2014-2024) with weather, socio-economic, infrastructure, and seasonal event data. Incorporating social and demographic indicators reveals underlying patterns of community vulnerability and provides a clearer understanding of outage risk during extreme conditions. Four machine learning models (Random Forest (RF), Support Vector Machine (SVM), Adaptive Boosting (AdaBoost), and Long Short-Term Memory (LSTM)) are evaluated. Experimental validation is performed on a large-scale dataset covering counties in the lower peninsula of Michigan. Among all models tested, the LSTM network achieves the lowest prediction error. Additionally, the results demonstrate that stronger economic conditions and more developed infrastructure are associated with lower outage occurrence.
Paper Structure (15 sections, 12 equations, 11 figures, 3 tables, 2 algorithms)

This paper contains 15 sections, 12 equations, 11 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Data Description and Preparation
  3. Historical Outage Data
  4. Historical Weather Data
  5. Socio-economic Data
  6. Power Infrastructure Data
  7. Data-driven Outage Prediction Algorithm
  8. Identifying Extreme Weather Event Data
  9. Identifying Extreme Weather Event Data
  10. Input Data Balancing
  11. Model Selection and Training
  12. Performance Evaluation
  13. Outage Prediction Model Formulation
  14. Experimental Setup and Simulation Results
  15. Conclusions

Figures (11)

  • Figure 1: Number of customers outage in Wayne County during 2024 based on EAGLE-1 dataset.
  • Figure 2: Weather stations locations in Lower Peninsula of Michigan.
  • Figure 3: Power infrastructure network of counties in Lower Peninsula of Michigan.
  • Figure 4: The framework of the proposed power outage prediction algorithm.
  • Figure 5: Number of Extreme Events by County (2014–2024).
  • ...and 6 more figures