Outage Prediction And Detecting Influential Factors Under Extreme Weather Events Using SHAP Analysis
Date of Award
December 2024
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Electrical Engineering
First Advisor
Prakash Ranganathan
Abstract
As electricity demands continue to grow and extreme weather events become more frequent, the need for resilient energy infrastructure is increasingly urgent. This thesis addresses three critical areas of power system resilience and monitoring. The first section of the thesis reviews the Texas power blackout in February 2021, which highlighted severe vulnerabilities in energy systems under extreme weather conditions. This analysis investigates the blackout event's causes, including generation capacity limitations and over-reliance on specific energy sources. The second section presents another weather event (ice storms) in North Dakota causing a multi-day blackout. Multiple machine learning approaches are deployed to predict power outages caused by these ice storms, using historical meteorological data. The study applies various algorithms to forecast outages based on features such as wind gust, wind direction, temperature, and precipitation metrics where the RFC model achieves the highest precision. The final section focuses on UAV-based inspection of high-voltage transmission lines, where electromagnetic interference (EMI) from electric and magnetic fields can disrupt UAV sensors. This study can aid during restoration of power lines using UAV as surveillance aid, but cognizant of E/H fields. Together, these studies offer valuable contributions to improving the resilience of power systems in the face of extreme weather and modernizing transmission line monitoring technologies through predictive models and UAVs.
Recommended Citation
Rahman, Farishta, "Outage Prediction And Detecting Influential Factors Under Extreme Weather Events Using SHAP Analysis" (2024). Theses and Dissertations. 6563.
https://commons.und.edu/theses/6563