一种暂态稳定并行仿真的改进算法及其加速比分析

提出了一种基于因子路径树网络划分的暂态稳定空间并行仿真的改进算法。采用新的网络划分性能评价指标、任务划分后续调整策略进行任务划分,使各处理器间计算负荷的分配更为合理;同时,还提出了该算法在理想状况下加速比的计算公式,分析了影响并行仿真计算效率的各个因素,并以某3872节点系统为例介绍了一种估计实现最大加速比所需处理器数目的方法。在Cluster1350集群系统上的算例表明,所提改进算法提高了仿真效率。

Improved parallel algorithm for transient stability simulation and analysis of its speedup

An improved parallel - in - space algorithm based on factorization path tree partitioning is proposed for transient stability simulation. To make the distribution of computing load among processors more reasonable,a new performance index for assessing the partitioning scheme and a post - adjusting strategy are adopted for the task assignment in this algorithm. A formula,which can be used to estimate the speedup under the ideal circumstance for this new algorithm is given,and the factors which influence the efficiency of the parallel simulation is analyzed. A method to evaluate the quantity of processors needed to achieve the maximum speedup is presented and is applied to a 3 872 - bus power system for test. The simulation results obtained on the Cluster 1 350 for the test power system validates that the improved algorithm has higher efficiency than the conventional one. This project is supported by National Natural Science Fund of China(50595412).