Data-Driven Volt/VAR Control of Power Distribution Networks

This paper proposes a data-driven control framework for voltage regulation in power distribution networks, eliminating the need for explicit system models. Instead, system dynamics are inferred from measurement data using linear subspace identification. Stability is ensured through a bilinear matrix inequality (BMI) formulation based on Lyapunov conditions, balancing theoretical rigour with practical feasibility. An adaptive power-sharing strategy enhances actuator coordination, facilitating decentralised volt/VAR regulation and improving system resilience. Extensive simulations on the IEEE 33-node test feeder under varying loads and high renewable penetration validate the approach, demonstrating enhanced stability, efficiency, and adaptability. By integrating data-driven control with theoretical guarantees, this work presents a scalable, model-free solution for modern power distribution networks.