Seismic precursors and operational earthquake forecasting impact on AI-based power network management

Seismic hazards, involving the cascading effects of earthquake-induced landslides and tsunamis, can impact infrastructure networks characterized by a complexity that is increased by the climate and energy transition. Novel management policies are needed to address the associated challenges. To achieve this, approaches leveraging "Artificial Intelligence" components are frequently proposed. However, their stochastic nature demands comprehensive and quantified verification and validation (V2) results. These results can, for example, describe the probability distribution of failures and identify the most probable and worst-case failure modes for the combined system of the controlled network and management system.  For geographically distributed systems, realistic V2 results must account for risks including those stemming from extreme events of seismic  origin.

We present a proof of concept demonstrator to validate management and control systems operating on a power network  in face of extreme events, integrating operational earthquake forecasting and seismic precursors maps.