提高预处理共轭梯度法计算大型电网潮流时并行性能的方法

Method for Parallel Performance Enhancement of Power Flow Computation for Large-Scale Power Grid by Use of Preconditioned Conjugate Gradient Method

研究了如何提高预处理共轭梯度(preconditioned conjugate gradient,PCG)法计算大网络交流潮流时的并行性能,提出了一种新的并行处理方法——并行节点分配法,即将节点导纳矩阵和节点出力的数据以节点为单位分派给各个处理器,在各个处理器中完成余下的计算处理部分。算例分析表明,在使用PCG法进行大网络交流潮流计算时,并行节点分配法在并行性能上具有一定的优势,进而为PCG法在大网络交流潮流并行计算中的应用提供了可借鉴的经验。

Research on how to improve parallel performance of AC power flow calculation for large-scale power grid by use of preconditioned conjugate gradient （PCG） method is conducted, and a new parallel processing approach, named as the parallel node distribution method, is proposed in which taking the node as unit the data of nodal admittance matrix and nodal output are assigned to each several processors and the remaining parts of calculation and processing are completed in these processors. The analysis and comparison results of various calculation examples show that when PCG method is used in AC power flow calculation of large-scale power grid, the proposed parallel node distribution method is dominant in parallel performance, thus it is available for reference in the application of PCG method in AC power flow calculation of large-scale power grid.