Project description
The research project AI-State-Estimation builds on the results of the preliminary project AI4Grids and further develops AI-based state estimation for distribution networks. While AI4Grids demonstrated the potential of combining load flow calculations with neural networks in a limited network area of Stadtwerke Kapfenberg, the follow-up project addresses the methodological and data-related challenges identified for large-scale application.
At the core of the project is the development of a hybrid approach based on Physics-Guided Neural Networks, enabling more accurate and real-time state estimation. By the end of the project, the state estimator is expected to reach a technology readiness level of TRL 7 and reliably detect at least 90% of critical network states while maintaining a false alarm rate below 10%. An improved data basis is achieved through additional measurement points in the network and the development of a load profile generator that converts network consumption and generation data of varying temporal resolution into high-resolution 15-minute profiles.
In addition, a recommendation engine is being developed that suggests adaptive network control measures based on the state estimation. The use of reinforcement learning enables predictive, self-learning network management, providing a significant contribution to the digitalization and flexibilization of distribution networks.
Project partner
- EnliteAI GmbH
- Stadtwerke Kapfenberg GmbH



