A Distributed-MPC Framework for Voltage Control Under Discrete Time-Wise Variable Generation/Load

Abstract

This article presents a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">distributed</i> predictive design for real-time voltage control under changing load and generation profiles at discrete intervals. The existing control designs include <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">centralized</i> approach, providing an optimal solution but less scalable and susceptible to single-point failures/attacks, as well as <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">decentralized or localized</i> approach, having increased scalability and attack resilience but lacking optimality. The proposed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">distributed</i> solution offers the attractive features of both approaches, where the neighboring nodes share their local information to attain an optimal solution while retaining scalability and resilience to single-point failures/attacks. We first introduce the centralized version of the voltage control problem assuming grid observability and then transfer it to the distributed versions based on both bus-wise and area-wise decompositions of the network. The distributed version is solved via alternating direction method of multipliers (ADMM) that, for bus-wise decomposition, needs a full set of local measurements, whereas only a partial set of local measurements (that guarantee area-wise grid observability for each area) is needed for area-wise decomposition, along with neighbor-to-neighbor communications. Additionally, leveraging the availability of measurement data, the framework includes a distributed method to estimate the admittance matrix <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\mathbf {Y}$</tex-math></inline-formula> of the underlying network graph. The proposed framework is validated against IEEE-30, IEEE-57 bus transmission systems, and IEEE-123 bus distribution systems and can tolerate certain levels of generation/load prediction uncertainties, modeling errors, and communication failures; plus, its in-built redundancy supports attack detection.