An Online Feedback Optimization Approach to Voltage Regulation in Inverter-Based Power Distribution Networks

Abstract

We address the problem of controlling the reactive power setpoints of a set of distributed energy resources (DERs) in a power distribution network so as to mitigate the impact of variability in uncontrolled power injections associated with, e.g., renewable-based generation. We formulate the control design problem as a stochastic optimization problem, which we solve online using a modified version of a projected stochastic gradient descent (PSGD) algorithm. The proposed PSGD-based algorithm utilizes sensitivities of changes in bus voltage magnitudes to changes in DER reactive power setpoints; such sensitivities are learned online via a recursive least squares estimator (rLSE). To ensure proper operation of the rLSE, the sequence of incremental changes in DER reactive power setpoints needs to be persistently exciting, which is guaranteed by a mechanism built into the controller. We analyze the stability of the closed-loop system and showcase controller performance via numerical simulations on the IEEE 123-bus distribution test feeder.