Distributionally Robust Safety Filter for Learning-Based Control in Active Distribution Systems

TL;DR

This paper introduces a distributionally robust safety filter that significantly reduces constraint violations in learning-based control of active distribution systems, ensuring safety during training without sacrificing near-optimality.

Contribution

It proposes a universal safety filter formulated as a distributionally robust optimization problem with chance constraints, applicable to any DRL agent in active distribution systems.

Findings

Reduces operational constraint violations during DRL training.

Guarantees constraint satisfaction with high probability.

Validated on IEEE 33-bus and 123-bus systems.

Abstract

Operational constraint violations may occur when deep reinforcement learning (DRL) agents interact with real-world active distribution systems to learn their optimal policies during training. This letter presents a universal distributionally robust safety filter (DRSF) using which any DRL agent can reduce the constraint violations of distribution systems significantly during training while maintaining near-optimal solutions. The DRSF is formulated as a distributionally robust optimization problem with chance constraints of operational limits. This problem aims to compute near-optimal actions that are minimally modified from the optimal actions of DRL-based Volt/VAr control by leveraging the distribution system model, thereby providing constraint satisfaction guarantee with a probability level under the model uncertainty. The performance of the proposed DRSF is verified using the IEEE 33-bus and 123-bus systems.