Scholarlink

Power outages resulting from severe weather conditions and increased energy demand have emerged as a critical issue for the stability and management of electrical grids. This study introduces a predictive system that employs probability-based statistical models alongside machine learning (ML) techniques, utilizing historical data on weather patterns and electricity usage. The primary aim is to pinpoint key factors that affect the likelihood of outages and to develop predictive models that can forecast potential disruptions to the grid. The system integrates logistic and Poisson regression with ML methods, including Random Forest (RF) and Support Vector Machines (SVM). The performance of these models is assessed through both historical data and simulated extreme scenarios. The RF model, which demonstrated the highest performance, achieved a prediction accuracy of 93%. Evaluation criteria encompass the accuracy of customer notifications, reduction in economic losses, savings in repair time, and enhancements in grid resilience. This research illustrates that employing data-driven predictive modeling can significantly improve outage management strategies and mitigate the adverse effects of power interruptions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predicting Power Outage Probabilities Using Weather and Consumption Data with Probabilistic Methods and Machine Learning

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Esra Dolgun,
  • Ibraheem Shayea,
  • Abdulraqeb Alhammadi,
  • Leila Rzayeva

摘要

Power outages resulting from severe weather conditions and increased energy demand have emerged as a critical issue for the stability and management of electrical grids. This study introduces a predictive system that employs probability-based statistical models alongside machine learning (ML) techniques, utilizing historical data on weather patterns and electricity usage. The primary aim is to pinpoint key factors that affect the likelihood of outages and to develop predictive models that can forecast potential disruptions to the grid. The system integrates logistic and Poisson regression with ML methods, including Random Forest (RF) and Support Vector Machines (SVM). The performance of these models is assessed through both historical data and simulated extreme scenarios. The RF model, which demonstrated the highest performance, achieved a prediction accuracy of 93%. Evaluation criteria encompass the accuracy of customer notifications, reduction in economic losses, savings in repair time, and enhancements in grid resilience. This research illustrates that employing data-driven predictive modeling can significantly improve outage management strategies and mitigate the adverse effects of power interruptions.