Machine Learning Application to Extreme Weather Power Outage Forecasting in Distribution Networks using a Majority Under-Sampling and Minority Over-Sampling Strategy

In the race against climate change as the frequency and severity of extreme weather events increases power system contingency analysis becomes increasingly critical. Power system operators can leverage Machine Learning (ML) in emergency response planning and operation to minimize the risks associate...

Full description

Saved in:
Bibliographic Details
Published in2023 IEEE Power & Energy Society General Meeting (PESGM) pp. 1 - 6
Main Authors Bahrami, Anahita, Shahidehpour, Mohammad, Pandey, Shikhar, Nation, Will, DSouza, Keith, Zheng, Honghao
Format Conference Proceeding
LanguageEnglish
Published IEEE 16.07.2023
Subjects
Ice
Ice Storm
Logistic regression
Machine learning
Majority Under-Sampling and Minority Over-Sampling
Minimum Redundancy Maximum Relevance
Power Distribution Grid Resilience
Prediction algorithms
Predictive models
Storms
Termination of employment
Online AccessGet full text

Cover

More Information
Summary:In the race against climate change as the frequency and severity of extreme weather events increases power system contingency analysis becomes increasingly critical. Power system operators can leverage Machine Learning (ML) in emergency response planning and operation to minimize the risks associated with power outages caused by extreme weather events. This paper presents a Machine Learning (ML) based classifier based on logistic regression theory to predict power distribution grid outages caused by an ice storm. The proposed Majority Under-Sampling and Minority Over-sampling (MUMO) addresses the frequency bias commonly present in power system outage data and improves the outage prediction model performance through Minimum Redundancy Maximum Relevance (MRMR) feature engineering and cross-validation.
ISSN:1944-9933
DOI:10.1109/PESGM52003.2023.10252804