基于Beowulf集群的大电力系统可靠性评估蒙特卡罗并行仿真  

The Parallel Monte-Carlo Simulation of Bulk Power System Reliability Evaluation Based on Beowulf Cluster

作　　者：刘洋[1] 周家启[1] 谢开贵[1] 胡小正[2] 程建翼[3] 曾伟民[3] 赵渊[1] 陈炜骏[1] 胡博[1] 

机构地区：[1]重庆大学高电压与电工新技术教育部重点实验室,重庆市沙坪坝区400044 [2]中国电力企业联合会可靠性管理中心,北京市宣武区100761 [3]湖北省电力公司,湖北省武汉市430077

出　　处：《中国电机工程学报》2006年第20期9-14,共6页Proceedings of the CSEE

基　　金：国家自然科学基金项目(50577072)~~

摘　　要：研究了大电力系统可靠性评估并行仿真问题:分别采用状态采样法和系统状态转移采样法进行Monte-Carlo并行仿真。建立了与收敛判据相结合的任务分配拓扑结构,并根据2种采样方法的不同需要采用了不同的伪随机数生成方式,详细地分析了配合收敛控制的异步模拟过程,最后基于构建的Beowulf集群环境进行测试系统的可靠性评估计算。2种Monte-Carlo并行仿真方法均得到较高的加速比和并行效率,其可靠性指标的计算结果亦与串行环境下得到的结果基本保持一致。该文所做工作是对国外若干电力系统可靠性评估并行仿真研究的深入和发展。This paper studies parallel simulation of bulk power system reliabilty evaluation. Using state sampling approach and system state transition sampling approach, two parallel Monte-Carlo simulation mothodologies are presented, including： the different topologies combining convegency check for two approaches are constructed, the random number sampling strategies for parallel processing are improved, and the asynchronous implementations with convegency control are analyzed in detailed. Finanily, the parallel methodologies are implemented on Beowulf cluster, the results obtained in tests with two power system models have the approximately same reliability indices with those in series processing, and show hign speed-up and parallel efficiency.

关 键 词：电力系统 可靠性评估 蒙特卡罗仿真 并行处理 BEOWULF集群 

分 类 号：TM74[电气工程—电力系统及自动化]