Scalable Congestion Forecasting for Electricity Grids (PredictiveClustering2Pythia)
This project integrates predictive clustering into the Pythia platform to improve and scale congestion forecasting for low-voltage electricity distribution grids.
The project has successfully enhanced the Pythia grid congestion forecasting pipeline through the integration of predictive clustering technology. The system has been validated using 2022 data from 43,000 digital meters, demonstrating significant performance improvements in forecasting load on low voltage feeders. Specifically, the new implementation achieved a 6% average and 13% median improvement in Continuous Ranked Probability Score (CRPS) compared to previous methods. The enhanced Pythia platform now automatically adapts to behavioral changes in electricity consumption driven by energy transition and local policies, reducing dependency on domain expert input
Contact: Koen Vanthournout (VITO), Wim Michielsen (VITO), Tim Op ’t Eyndt (VITO)
Researchers: Koen Vanthournout (VITO - Senior researcher)
Research groups: VITO W-ET unit