Predict. Optimize. Revise. On Forecast and Policy Stability in Energy Management Systems

  • 2025-02-03 17:41:20
  • Evgenii Genov, Julian Ruddick, Christoph Bergmeir, Majid Vafaeipour, Thierry Coosemans, Salvador Garcia, Maarten Messagie
  • 0

Abstract

This research addresses the challenge of integrating forecasting andoptimization in energy management systems, focusing on the impacts of switchingcosts, forecast accuracy, and stability. It proposes a novel framework foranalyzing online optimization problems with switching costs and enabled bydeterministic and probabilistic forecasts. Through empirical evaluation andtheoretical analysis, the research reveals the balance between forecastaccuracy, stability, and switching costs in shaping policy performance.Conducted in the context of battery scheduling within energy managementapplications, it introduces a metric for evaluating probabilistic forecaststability and examines the effects of forecast accuracy and stability onoptimization outcomes using the real-world case of the Citylearn 2022competition. Findings indicate that switching costs significantly influence thetrade-off between forecast accuracy and stability, highlighting the importanceof integrated systems that enable collaboration between forecasting andoperational units for improved decision-making. The study shows that committingto a policy for longer periods can be advantageous over frequent updates.Results also show a correlation between forecast stability and policyperformance, suggesting that stable forecasts can mitigate switching costs. Theproposed framework provides valuable insights for energy sector decision-makersand forecast practitioners when designing the operation of an energy managementsystem.

 

Quick Read (beta)

loading the full paper ...