Real‐Time Control of Electrical Distribution Grids

on YouTube ↗31m

" summary

A real-time control system called Comelec orchestrates inverters, batteries and PV by a central grid agent polling device-side resource agents every 100 ms; resources advertise a PQ profile (feasible active/reactive intervals), a virtual cost and a belief function (set of possible outcomes). The grid agent takes gradient steps to minimize total cost while enforcing admissibility via V‑control tests (robust feasibility against belief sets). Two pilot grids run now with four more deploying.

AI-written summary · about

" chapters (4)

1. 0:00Start
2. 2:351. Real-Time Operation of Microgrid: Motivation Absence of inertia (inverters) Stochastic generation (PV) Storage, demand response Grid stress (charging stations, heat pumps) Support main grid (primary and secondary frequency support)
3. 5:46Every 100 msec Grid Agent monitors grid and sends power setpoints to Resource Agents Resource agent sends to grid agent: PQ profile, Virtual Cost and Belief Function Goal: manage quality of service in grid; support main grid; use resources optimally [Bernstein et al 2015, Reyes et al 2015]
4. 9:06Belief Function Say grid agent requests setpoint (PsetOset) from a resource; actual setpoint (P.) will, in general, differ. Belief function exported by resource agent means: the resource implements (PQ) E BF (Pset, Oset)