基于电网运行大数据的在线分布式安全特征选择

简述大数据环境下,电网安全特征选择的现状与问题。提出了一种基于电网特征量相关性分组、适应于电网运行大数据的在线分布式安全特征选择方法,该方法能在线挖掘出关键的电网安全运行特征。首先阐述了单个计算节点上电网安全特征选择方法,接着提出了基于电网特征量分组的分布式安全特征选择方法;由于电网特征量分组情况会对特征选择结果产生较大影响,故提出了基于电网特征量相关性分组的策略,尽量使得同一组内的电网特征量相关性较大,不同分组间的电网特征量相关性较小。IEEE 9节点系统和广东实际省网系统算例验证了该方法的实用性和有效性,表明了该方法能够快速挖掘出电网运行的薄弱点,帮助电网运行人员准确地把握电网安全运行特征,同时也对比了该方法相比传统方法在计算准确性和计算速度方面的优势。

Online Distributed Security Feature Selection Based on Big Data in Power System Operation

The latest development and existing problems of power system security feature selection are briefly introduced. An online distributed security feature selection method is proposed. The method is based on power system security feature grouping by correlation and adapts to the big data in power system operation, and it could online discover critical features for power system security. First, a power system security feature selection method for a single compute node is discussed. Then a distributed method based on feature grouping is proposed. As the feature grouping method has an important influence on the selection results, a strategy based on power system security feature grouping by correlation is put forward to make correlation of features within the same group larger while the correlation of features among different groups smaller. This distributed security feature selection method is well applied in IEEE 9-bus system and Guangdong power system for its practicality and effectiveness, which could quickly find out the weak spots in power system operation and accurately help operators grasp the critical features for power system security. Compared with traditional methods, this method performs well for its compute accuracy and speed.