基于海量在线历史数据的大电网快速判稳策略  

Strategy of Huge Electric Power System Stability Quick Judgment Based on Massive Historical Online Data

作　　者：黄彦浩[1] 于之虹[1] 史东宇[1] 周孝信[1] 

机构地区：[1]中国电力科学研究院,北京市海淀区100192

出　　处：《中国电机工程学报》2016年第3期596-603,共8页Proceedings of the CSEE

基　　金：国家电网公司科技项目(XT71-14-04)~~

摘　　要：基于已有在线历史数据进行电力系统稳定性快速判断,可用于生成在线计算的故障列表,补充动态安全分析的故障集,使之在有限时间内考虑更多的故障情况。该文根据大系统运行特点,提出仅使用静态物理量的特征量选取方法并引入了更具稳健性的统计指标。针对实际系统数据失稳样本过少的问题,基于支持向量机(support vector machine,SVM)提出了"扩展边界"策略。根据电力系统运行的周期性,提出了针对海量历史数据的训练样本集构建策略,即按日选取故障前有限时间窗口的历史数据。综合上述研究成果,完成了快速判稳策略,给出了策略的实现流程。采用实际大系统数据,对提出的模型和策略进行了测试。结果表明,该文成果与大系统在线计算快速判稳的要求相切合,具有较好的性能和实际应用价值。Based on historical online data the electric power system stability quick judgment can be realized and used to make the additional fault list of online dynamic security analysis（DSA） for expending the considering fault situations. On the base of huge electric power system operation characteristics, the method of feature quantities selection is proposed which only includes quiescent physical quantities and uses robust statistic technology. For solving the problem of seldom stability samples, the expanding boundary method is proposed by using support vector machine（SVM）. Based on the periodicity of huge electric power system operation, the sample set building method according to massive historical online data is proposed which selecting the historical data by the limited number of days before the fault. Based on all of above achievements, the stability quick judgment strategy is realized and the strategy flow is given. The historical online data of an actual huge electric power system is used to test the strategy. The results show that the achievements of this paper fit to the requirements of huge electric power system online stability quick judgment and have good performance and practical application value.

关 键 词：快速判稳 在线数据 支持向量机 稳健统计量 

分 类 号：TM712[电气工程—电力系统及自动化]