FedPPO-PG Enhances Smart Grid Stability with Distributed RL

Omar Al-Refai, Ibrahim Shahbaz, Adam Ali Husseinat, Eman Hammad· July 8, 2026 View original

Summary

FedPPO-PG is a Federated Multi-Agent Proximal Policy Optimization framework that improves transient stability control in smart grids by reformulating it as a cooperative multi-agent reinforcement learning problem. It uses physics-grounded neighborhoods and a centralized training-decentralized execution paradigm to achieve rapid post-fault damping.

Maintaining transient stability in smart grids is crucial to prevent cascading failures, requiring rapid damping of generator frequency and rotor angle deviations after a fault. This paper introduces FedPPO-PG, a novel Federated Multi-Agent Proximal Policy Optimization framework designed to address this challenge. It redefines transient stability control as a cooperative multi-agent reinforcement learning problem, directly optimizing against closed-loop stability objectives. Each generator in the system hosts an independent local actor, augmented with frequency deviations from its two most strongly coupled electrical neighbors, identified through a physics-grounded approach. The framework employs a centralized training-decentralized execution (CTDE) paradigm, where a centralized critic guides advantage estimation, while actors operate independently at deployment. Evaluated on the IEEE 39-bus system, FedPPO-PG achieved 100% stabilization, significantly reduced stability time, and cut control power compared to baselines, all while meeting real-time inference latency requirements.

Why it matters

For energy sector professionals and grid operators, this research offers a highly effective, distributed, and real-time capable AI solution for enhancing smart grid stability, preventing blackouts, and optimizing power control.

How to implement this in your domain

  1. 1Evaluate the current transient stability control systems in smart grids for potential integration of AI-driven solutions.
  2. 2Pilot FedPPO-PG or similar federated reinforcement learning frameworks in simulated smart grid environments.
  3. 3Collaborate with energy researchers to adapt physics-grounded AI models for specific grid architectures and operational constraints.
  4. 4Develop real-time data acquisition and communication infrastructure to support decentralized AI control agents.

Who benefits

EnergyUtilitiesInfrastructureGovernmentAI/ML Platforms

Key takeaways

  • FedPPO-PG significantly improves transient stability control in smart grids.
  • It uses a federated multi-agent RL approach with physics-grounded insights.
  • The framework achieves 100% stabilization and reduces control power and stability time.
  • Actors operate independently at deployment, meeting real-time requirements.

Original post by Omar Al-Refai, Ibrahim Shahbaz, Adam Ali Husseinat, Eman Hammad

"arXiv:2607.05553v1 Announce Type: new Abstract: Transient stability control in smart grids requires rapid post-fault damping of generator frequency and rotor angle deviations to prevent cascading failures. This paper proposes FedPPO-PG, a Federated Multi-Agent Proximal Policy Opt…"

View on X

Originally posted by Omar Al-Refai, Ibrahim Shahbaz, Adam Ali Husseinat, Eman Hammad on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses