Research launchedResilienceNetwork restorationNetwork resilience
The approach
RESTORE stands for RESilient Topology Optimization for REcovery of Networks. The project explores methods to restore infrastructure networks after a major disruption.
The work combines reinforcement learning and physics-informed neural networks. Mario Real Enrique leads the doctoral research, with Sara Ghane in the scientific direction on the company side.
Launched in 2026 and planned until 2030, the project is led by DTAS with the University of Antwerp, under a Baekeland mandate supported by VLAIO.
Objectives
- Physics-informed, topology-aware AI that predicts cascading disruptions in critical networks and recommends fast, safe recovery actions.
- Predictor: spatio-temporal heterogeneous GNN world model. Actor: risk-aware multi-agent reinforcement learning with a safety layer.
- Validation planned in two domains: energy (voltage stability) and supply chain (service levels). Target TRL 2 to 5.
The project team
- Mario Real EnriquePhD researcher, DTAS
- Dr Sara Ghaneco-promotor, DTSC
- Prof. Peter Hellinckxacademic promotor, UAntwerpen
- Dr Renzo Massobrioacademic promotor, UAntwerpen
The contributions
Source: DTSC Information · November 2026.
Objectives and team: the laboratory’s “BRAIN projects” document, version of 13 November 2026.
Discuss this project
