Towards an AI Assistant for Tertiary Voltage Control
summary
An AI assistant uses PyPowSybl and graph models to run on a full 7K buses French grid with no preprocessing, cutting over-voltage events by roughly half via unsupervised tertiary voltage control; contingency screening flags 95% of risky outages with ~1,000 false positives and reduces compute time by ~8×; topology optimization on a 12-substation test increases transfer capacity by 9.3% in ~200 ms versus 13.7% for a classical solver that can take >10 minutes, highlighting scalability and robustness challenges.
AI-written summary · about
chapters (9)
- 0:00Introduction and motivation
- 1:07Why graph neural networks
- 3:36Handling topological variations
- 5:39Tertiary voltage control use case
- 7:50Contingency screening use case
- 10:18Topology optimization use case
- 12:45Argo framework overview
- 16:41Roadmap and future work
- 19:39Open-source release and Q&A
related talks
09mAINETUS - AI for Safety-Critical Network Infrastructures
17mAI Assisted Load Sharing for Improved Energy Efficiency
21mApplication of AI technologies in nuclear power plants
31mAI-supported transformation of the urban infrastructure
20mFrom Hype to Value: How AI is transforming the utilities industry
22mHow To Orchestrate RL Agents for Topological Remedial Actions Using PowSyBl’s Ope...
32mPredicting Grid Faults Before They Happen: AI for Weather-Driven Grid Resilience
18mFrom Smart Homes to Smart Grids