Conference paper

Edge AI Framework for Real-Time Distributed Energy Resource and Virtual Power Plant Integration

Authors
Kabir Olatoke and Moein Manbachi
Event
CIGRE Energy Forum 2026, Calgary, Alberta, Canada
Study Committee
C6, Active Distribution Systems and Distributed Energy Resources
Topics
Grid-enhancing techniques and technologies · Distribution management
Paper ID
10288
Affiliation
British Columbia Institute of Technology (BCIT), with VPP Dynamics Inc.

Summary

Rooftop solar, batteries and other inverter-based resources are growing fast on distribution grids. Control platforms hosted in the cloud must send every decision over a wide-area network, which makes it hard to meet the fast, local response that interconnection standards such as IEEE 1547-2018 and IEEE 2030.14 expect.

The paper describes an edge AI framework that moves time-critical decisions to embedded controllers close to the equipment. It follows a three-layer design: a device layer that talks to inverters over SunSpec Modbus, an edge layer that runs short-term solar forecasting, power-flow checks and Volt-VAR control, and a cloud layer that supervises the fleet as a virtual power plant.

The framework also separates the voltage used for ride-through protection from the voltage used for reactive power control, which helps avoid unnecessary protective disconnections. It was built and tested on a physical hardware-in-the-loop testbed running continuously, and the edge control kept working even when the cloud connection was slow.

The authors conclude that edge control gives utilities a practical way to let distributed energy resources take part in grid services without relying on fast wide-area communications.

Keywords: distributed energy resources, edge computing, hardware-in-the-loop, IEEE 1547, real-time control, virtual power plant, Volt-VAR control.

Request a copy

This summary is written by VPP Dynamics in plain language. The full paper, including test results, is shared on request. Once the conference proceedings are published, we will link to the official copy.

See how it works for your fleet.

Book a demo with our team, or join our BC, Alberta and Ontario pilot, launching in 2027.