Low-Voltage Grid Control Based on Data-Driven State Estimation and Online Feedback Optimization

The operation of low-voltage distribution grids is facing increasingly complex challenges with the integration of Distributed Energy Resources (DERs). To ensure secure grid operation, Distribution System Operators can utilize active network management with autonomous monitoring and control systems. In this paper, we propose a novel two-stage controller for LV grids that combines data-driven state estimation and model-free Online Feedback Optimization (OFO). The first stage is a data-driven state estimator that leverages real-time measurements at the substation and historical smart meter data to compute real-time voltage estimates at smart meter nodes. The second stage is an OFO algorithm fed with the voltage estimates obtained in the first stage. The OFO algorithm is designed to mitigate voltage band violations while minimizing DER intervention actions. Simulation experiments conducted on a benchmark LV grid with high DER penetration and realistic load and generation profiles demonstrate the ability of the proposed controller to mitigate both undervoltage and overvoltage situations while maintaining DER control interventions at a minimum.