Machine learning improves forecasting of german power grid interventions
Machine Learning for German Redispatch Forecasting under Data Delays and Temporal Distribution Shift
Machine Learning
Summary
Predicting energy grid interventions can be tricky because the data arrives late and includes many zero events. The authors tested different machine learning methods to forecast these interventions in Germany, even when information is delayed by at least a week. They found that certain models, especially boosted decision trees with careful adjustments, give more accurate predictions overall. However, the models sometimes underestimated extreme events when unusually large interventions occurred.
What this means in practice
- •For power grid operators: Produce reliable probabilistic forecasts of intervention energy volumes ahead of time despite data delays to support congestion management decisions.
- •For energy market analysts: Enhance short-term congestion forecasting models to better anticipate grid stress periods with improved uncertainty estimates using machine learning with delayed data.
Authors
Faraz Shamim, Faris Shamim
Abstract
Public redispatch records provide empirical data for grid congestion forecasting, but delayed reporting, zero-inflated distributions, and temporal shift present major modeling challenges. We assess the accuracy and reliability of probabilistic machine-learning forecasts using published German transmission records under experimentally imposed information-age constraints. The benchmark evaluates eight daily series of upward and downward intervention energy across four German transmission system operators from 2021 to 2024 (48,242 eligible records; 354 evaluation dates in 2024). We compare seasonal empirical, regularized autoregressive (ARX), quantile LightGBM, GRU, and Transformer models under a minimum seven-day target-latency constraint. Neural architectures use a zero-censored output head to accommodate exact-zero outcomes. Static, rolling, and adaptive delayed-feedback calibration are evaluated using normalized weighted interval score (nWIS), empirical coverage, and block-bootstrap inference. Raw LightGBM achieved nWIS 0.7952, outperforming ARX (1.0604) and the seasonal baseline (0.8739) by 25.0% and 9.0%, respectively (Holm-adjusted p<0.005). Rolling calibration improved LightGBM to nWIS 0.7767 versus 0.8251 for static calibration (p=0.0092), with 91.81% coverage for nominal 90% intervals. The zero-censored Transformer achieved nWIS 0.8161, with no significant difference from LightGBM (p=0.260). However, aggregate coverage concealed substantial undercoverage during high-volume interventions (61.91% coverage among above-threshold events). These results show that boosted-tree models with rolling calibration provide accurate probabilistic forecasts of aggregate redispatch volumes under target delays, while nominal aggregate validity does not ensure reliability during extreme congestion events.