Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Probabilistic forecasting of weather-driven faults in electricity networks: a flexible approach for extreme and non-extreme events

Mateus Maia, Daniela Castro-Camilo, Jethro Browell

TL;DR

This work presents a novel probabilistic framework for forecasting fault counts that captures typical and extreme events, and uses ensemble numerical weather predictions to incorporate the impact of weather fluctuations.

Abstract

Electricity networks are vulnerable to weather damage, with severe events often leading to faults and power outages. Timely forecasts of fault occurrences, ranging from nowcasts to several days ahead, can enhance preparedness, support faster response, and reduce outage durations. To be operationally useful, such forecasts must quantify uncertainty, enabling risk-informed resource allocation. We present a novel probabilistic framework for forecasting fault counts that captures typical and extreme events. Non-extreme faults are modeled linearly interpolating estimates from multiple additive quantile regressions, while extreme events are described through a discrete generalized Pareto distribution. To incorporate the impact of weather fluctuations, we use ensemble numerical weather predictions, which helps to quantify uncertainty in the forecasts. This approach is designed to provide reliable fault predictions up to four days ahead. We evaluate the model through numerical experiments and apply it to historical fault data from two electricity distribution networks in Great Britain. The resulting forecasts demonstrate substantial improvements over business-as-usual and alternative modeling approaches. A practitioner trial conducted with Scottish Power Energy Networks from October 2024 to March 2025 further demonstrates the operational value of the forecasts. Engineers found them sufficiently reliable to inform decision-making, offering benefits to both network operators and electricity consumers.

Probabilistic forecasting of weather-driven faults in electricity networks: a flexible approach for extreme and non-extreme events

TL;DR

This work presents a novel probabilistic framework for forecasting fault counts that captures typical and extreme events, and uses ensemble numerical weather predictions to incorporate the impact of weather fluctuations.

Abstract

Electricity networks are vulnerable to weather damage, with severe events often leading to faults and power outages. Timely forecasts of fault occurrences, ranging from nowcasts to several days ahead, can enhance preparedness, support faster response, and reduce outage durations. To be operationally useful, such forecasts must quantify uncertainty, enabling risk-informed resource allocation. We present a novel probabilistic framework for forecasting fault counts that captures typical and extreme events. Non-extreme faults are modeled linearly interpolating estimates from multiple additive quantile regressions, while extreme events are described through a discrete generalized Pareto distribution. To incorporate the impact of weather fluctuations, we use ensemble numerical weather predictions, which helps to quantify uncertainty in the forecasts. This approach is designed to provide reliable fault predictions up to four days ahead. We evaluate the model through numerical experiments and apply it to historical fault data from two electricity distribution networks in Great Britain. The resulting forecasts demonstrate substantial improvements over business-as-usual and alternative modeling approaches. A practitioner trial conducted with Scottish Power Energy Networks from October 2024 to March 2025 further demonstrates the operational value of the forecasts. Engineers found them sufficiently reliable to inform decision-making, offering benefits to both network operators and electricity consumers.
Paper Structure (30 sections, 7 equations, 19 figures, 5 tables)

This paper contains 30 sections, 7 equations, 19 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Quantile Regression
  4. Discrete generalized Pareto distribution
  5. Bayesian Model Averaging for Weather and Fault Forecasting
  6. Methodology
  7. Model Specification
  8. Estimation and Forecast
  9. Implementation & Benchmarking
  10. Simulation Experiments
  11. First scenario: scale-invariance in the tail distribution
  12. Second scenario: non-stationary in the tail distribution
  13. Case Study: short-term fault forecast for Scotland's electrical networks
  14. Data description
  15. Network and Fault Data
  16. ...and 15 more sections

Figures (19)

  • Figure 1: Daily occurrence of faults in a SPEN district over ten years (left), with the vertical dashed line indicating the red--event threshold. The right panel shows estimated probabilities of issuing a red warning on non-event days when exceeds 9m/s, comparing the naive and proposed models.
  • Figure 2: UK map of the two licence areas operated by : SP Distribution (SPD), covering Central and Southern Scotland (top, in green), and SP Manweb (SPM), covering North Wales, Merseyside, and Cheshire (bottom, in purple). District boundaries - spatial units over which faults are aggregated for modeling and prediction - are shown within license areas. Each district comprises several hundred circuits.
  • Figure 3: Reliability of X-flexForecast models for 24-hour resolution forecast, out-of-sample hindcasts using ERA5 for the windiest 20% of days.
  • Figure 4: 24-hour resolution results across all districts. Left panel: AUC ratio $>1$ indicates improvement over qgam-only approach. Right panel: Values $>$0 show X-flexForecast improvement in predicting category display probabilities. Dashed lines represent no change in performance.
  • Figure 5: Reliability of qgam models, out-of-sample re-forecast with lead-times from zero to four days-ahead using the ECMWF, and/or HRES deterministic forecast for the windiest 20% of days. The color for each district is the same as Figure \ref{['fig:qgam_model_rel']}.
  • ...and 14 more figures