A Fine-Grained Parallel Power Flow Method for Large Scale Grid Based on Lightweight GPU Threads

Résumé fourni par la source

This paper proposes a parallel Newton-Raphson Power Flow (PNPF) method which is suitable for GPU (Graphics Processing Unit). Aiming at the most time-consuming part of power flow-solving the sparse linear equations, an efficient hierarchy parallel solver is presented, in which LU decomposition and forward/back substitution were implemented in parallel level by level under the direction of path tree. To improve the efficiency of the method, a short-type path tree was formed, an optimization on data transfer and a self-adapted task-allocation algorithm was designed. In addition, the sparse Jacobian matrix and the right-hand side of the linear equations were also generated in parallel in a fine-grained pattern. At last, the developed GPU-based PNPF program has been tested on large-scale power systems of up to 23215 buses. The method provides a speedup of 3.91 times compared to mainstream commercial software, which proves the effectiveness and practicality of the algorithm proposed in this paper.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.