SENOM is an operational toolset that turns short-term forecasts into concrete, consequence-aware actions for MV/LV grids. It lets operators load grid data, run asset-level load/production forecasts, quantify alarm significance (voltage/current), and compare reconfiguration actions on interactive maps—directly from a browser-based UI.
Key features:
- Guided data prep: Upload/validate states, tags, hourly timelines, and geometries; optional assets (transformers, substations, DG). Consistent naming and health checks are built in.
- Live prediction: Inference UI for trained hybrid Transformer-KAN forecasters (separate heads for loads and DGs), with CPU/GPU support, plotting, and CSV export.
- Alarm significance explorer: Visualizes observed vs. predicted significance for current/voltage using a modified Chernoff-Bound, including fixed 72-hour evaluation windows.
- Prescription module: Compares hour-specific sets of pre-vetted radial switching configurations (top 20) and recommends actions via a PPO policy trained in a pandapower environment.
- Grid visualization: Map views with line-loading ramps, bus markers, CB-aware styling, time selectors, and basemap controls.
How it works:
SENOM consumes 1-hour-ahead forecasts, runs balanced power flow to obtain voltages/currents, ranks deviations via consequence-aware significance (scaled by affected consumers), and selects a switching action from feasible radial topologies using PPO for fast, robust decision-making.
Related publications:
Principal investigator: Hamid Reza Shaker
Development team: Hamid Mirshekali, Mohammad Reza Shadi
Tool link: Book a Demo (send email to hrsh@mmmi.sdu.dk)