Abstract

Managing a power grid requires constantly solving unknown node voltages and currents. Modified Nodal Analysis (MNA) is used to write Kirchhoff’s Current Law (KCL) at every node, resulting in a massive system of linear equations that are computationally expensive for dense classical solvers. Quantum methods, specifically the Variational Quantum Linear Solver (VQLS) algorithm, have been proposed to solve linear systems more efficiently. However, NISQ (Noisy Intermediate-Scale Quantum) processors do not have enough stable qubits or low error rates to process a massive global grid matrix. Furthermore, variational methods suffer from high execution latencies. In this talk, we will discuss our proposed method for solving linear systems in power grid simulation, using distributed VQLS agents and Field-Programmable Gate Array (FPGA) accelerated optimization and coordination. The power-network problem is formulated as a real-valued MNA system and decomposed into coupled sub-grids using Diakoptics (domain decomposition). Each sub-grid is assigned to an independent quantum agent that solves a local VQLS problem and produces boundary-voltage information for a global coordinator. A coordination layer implementing the Alternating Direction Method of Multipliers (ADMM) updates boundary targets and variables to enforce tie-line consistency across the decomposed network. A prototype implementation was evaluated using the IEEE 9-bus benchmark. Obtained simulation results validate the feasibility of the Diakoptics-based, multiple quantum agent partitioning scheme and the heterogeneous framework.

About the Speaker

Naveed Mahmud is an assistant professor in the Department of Electrical Engineering and Computer Science (EECS) at Flroida Institute of Technology. He received his Ph.D. degree in Electrical Engineering from the University of Kansas, USA, in 2022. His primary research interests are computer architecture, reconfigurable computing, quantum computing, quantum communications, and heterogeneous computing. His research work is funded by organizations such as the National Science Foundation (NSF). His teaching interests are Computer Architecture, Programmable Gate Arrays, and Electrical Circuit Theory.