CommonPower: A Framework for Safe Data-Driven Smart Grid Control

TL;DR

CommonPower addresses the challenge of safely deploying data-driven controllers in smart grids by providing a symbolic-model-based framework that automatically synthesizes robust MPC and RL safeguards. It offers modular modeling, an integrated forecasting workflow, and a unified interface for single- and multi-agent RL, enabling flexible deployment across decentralized power systems. The paper demonstrates robustness, scalability, and the influence of forecaster choice through experiments in building and microgrid scenarios, highlighting practical implications for real-world adoption. The framework thus facilitates rapid, safe case-study design and comparative evaluation of heterogeneous control strategies in future energy systems.

Abstract

The growing complexity of power system management has led to an increased interest in reinforcement learning (RL). To validate their effectiveness, RL algorithms have to be evaluated across multiple case studies. Case study design is an arduous task requiring the consideration of many aspects, among them the influence of available forecasts and the level of decentralization in the control structure. Furthermore, vanilla RL controllers cannot themselves ensure the satisfaction of system constraints, which makes devising a safeguarding mechanism a necessary task for every case study before deploying the system. To address these shortcomings, we introduce the Python tool CommonPower, the first general framework for the modeling and simulation of power system management tailored towards machine learning. Its modular architecture enables users to focus on specific elements without having to implement a simulation environment. Another unique contribution of CommonPower is the automatic synthesis of model predictive controllers and safeguards. Beyond offering a unified interface for single-agent RL, multi-agent RL, and optimal control, CommonPower includes a training pipeline for machine-learning-based forecasters as well as a flexible mechanism for incorporating feedback of safeguards into the learning updates of RL controllers.

Figures (9)

• Figure 1: Microgrid of a university campus with several prosumers and different control architectures that can be realized using CommonPower.
• Figure 2: UML class diagram of power system entities.
• Figure 3: UML class diagram of the data provider structure.
• Figure 4: Simplified visualization of one simulation step, highlighting the modularity in CommonPower.
• Figure 5: Control flow during training and deployment in CommonPower.