Symbolic-AI driven Data Repairs for Large Scale Energy Co-Simulations: Combining SHACL repairs and Datalog rules to detect, explain, and correct errors in large scale energy co-simulation setups
Published in: Posters and Demos Track of SEMANTiCS
2025
Abstract
The transformation of energy distribution systems is fostering new models, like renewable energy communities, which require complex, simulation-based feasibility assessments. Preparing these simulations is often labor-intensive and error-prone due to heterogeneous actors and location-specific grid topologies. This paper proposes a symbolic AI approach that combines SHACL (repairs) and Datalog (imputation) to semi-automatically detect, explain, and correct inconsistencies for grid and sensor data so it can serve as input for co-simulations. Applied within the DataBri-X project and tested using Siemens BIFROST, the approach demonstrates promising improvements in data quality and preprocessing efficiency.