A Fully Parallel and Scalable FDSE Approach Based on Balanced Measurements Partitioning and Graph Node-Centric Hash Storage

Parallel computing approaches have illustrated promising potential in accelerating state estimation (SE). How to efficiently handle data processing and updating has become a crucial problem in improving the performance of SE. This article proposes a fully parallel and scalable approach (BMPGNC) based on balanced measurements partitioning and graph node-centric hash (GNCH) storage, to accelerate fast decoupled state estimation (FDSE) for largescale power systems, which is based on graph model multithread parallelism without any centralized preprocessing. A GNCH model is presented for compressed data storage and fast data access, where a balanced measurement partitioning is utilized for load balancing over multiple threads. Four GNCH-based algorithms are proposed for parallel data processing and updating. Extensive experiments were conducted over six benchmark test systems. The results show BMPGNC achieves superior performance on accelerating FDSE without compromising high accuracy, thereby achieving fast, energy-efficient, and high-accuracy FDSE for large-scale power systems.