Accelerating Quantum Optimization with Graph Learning for Optimal PMU Placement

The Phasor Measurement Unit (PMU) plays a crucial role in the real-time monitoring of power systems. Investigating the optimal PMU placement can not only provide full system observability but also reduce unnecessary installation costs. However, as a typical combinatorial optimization problem, determining the optimal PMU installation strategy is NP-hard and would cost significant computational resources for classical computers. In this work, a graph learning based acceleration strategy for the Quantum Approximation Optimization Algorithm (QAOA) is developed to effectively identify the optimal PMU placement. The proposed graph learning method will distill knowledge from the power grid to better initialize the QAOA parameter, boosting the overall implementation speed of QAOA. Empirical results indicate that our graph learning strategy can provide a 5% to 8% speedup to the QAOA execution. This work aims to provide a cornerstone in accelerated quantum computing to solve real-world power system applications.