摘要
由于电网运行数据具有多源、异构、高维等典型大数据特征,使得传统检测方法已无法实现异常数据高效辨识;因此提出一种基于Spark框架的电网运行异常数据辨识与修正新方法。首先,提出了并行化最小生成树方法对待检测数据进行初始聚类;在此基础上结合并行K-means算法对数据进行二次聚类实现异常数据辨识;然后,在Spark框架下设计了基于径向基函数(RBF)神经网络的异常数据修正模型,实现对异常数据修正。最后,利用某省调度中心SCADA数据对方法的有效性进行了验证,结果表明所提方法能够有效处理电网运行异常数据,具有实际应用价值。
The operation data of power grid has the characteristics of multi-source,heterogeneous,high-dimensional and other typical big data,which makes it impossible for traditional detection methods to identify the abnormal data efficiently. Therefore,a new method of identifying and correcting the abnormal data of power grid was proposed based on Spark. First of all,used the parallel minimum spanning tree method was to cluster the detected data initially. On the basis of this,combined with parallel K-means algorithm for secondary clustering of data the abnormal data identification was realized. Then,an abnormal data correction model based on radial basis function( RBF) neural network is designed in the Spark framework to correct the abnormal data. Finally,the effectiveness of the method is verified by the SCADA data of a provincial dispatching center. The results show that the proposed method can effectively deal with the abnormal data of grid operation and has practical application value.
作者
曲朝阳
朱润泽
曲楠
曹令军
吕洪波
胡可为
QU Zhao-yang;ZHU Run-ze;QU Nan;CAO Ling-jun;L Hong-bo;HU Ke-wei(School of Computer Science of Northeast Electric Power University, Jilin 132012, China;Jilin Engineering TechnologyResearch Center of Intelligent Electric Power Big Data Processing, Jilin 132012, China;Maintenaue Company of JiangsuPower Company, Nanjing 210000, China;State Grid Jilin Electric Power Supply Company, Changchun 130000, China)
出处
《科学技术与工程》
北大核心
2019年第25期211-219,共9页
Science Technology and Engineering
基金
国家自然科学基金重点项目(51437003)
吉林省科技发展计划重点项目(20180201092GX)
吉林省科技发展计划(20160623004TC)资助
