Forecasting Electricity Loads with EDF

Since 2010, our department has collaborated with EDF Paris to develop advanced methods for electricity load forecasting, a critical component for efficient scheduling and decision-making in energy markets.

Early work focused on modelling daily load patterns using curve linear regression with SVD-based dimension reduction, capturing the mean level of future consumption. As energy systems evolve, with increased renewable integration and more volatile markets, extreme events and fluctuations have a growing impact on operational costs. This motivates probabilistic forecasting, providing not only accurate point predictions but also predictive intervals and quantiles that help manage risk and inform strategic decisions.