基于PCA谱聚类分析的无功分区方法研究  

Study of Reactive Power Control Partitioning Method with Spectral Cluster Analysis Based on PCA

作　　者：肖李明 周玲[1] 张小龙[1] 徐辉 杨在辅 

机构地区：[1]河海大学能源与电气学院,江苏南京211100 [2]国网江苏省电力公司检修分公司,江苏南京211100 [3]安徽理工大学电气与信息工程学院,安徽淮南233001

出　　处：《陕西电力》2016年第12期23-28,共6页Shanxi Electric Power

基　　金：国家自然科学基金项目资助(11575003)

摘　　要：以互联电网简化拓扑结构模型为基础,结合复杂网络理论,提出了一种基于PCA谱聚类分析的无功电压分区新方法。该方法根据电网的运行和结构特性,以网络的加权邻接矩阵作为无功分区的简化拓扑模型,利用PCA对分区模型进行主成分分析,将主成分个数作为谱聚类选择特征向量的维数标准,从而改进了谱聚类算法并完成对电力网络的快速有效划分。同时,把无功缺额度与模块度Q函数结合起来构建新的模块度指标,保证了分区数目的合理性并改善了区域的无功平衡性,最后对分区结果进行无功备用校验并做出相应调整从而得到最终的分区方案。通过对IEEE 30标准节点的仿真计算和测试,验证了该分区方法的有效性。On the basis of the simplified interconnection power grids topology model, combined with complex network theory, the paper puts forward a new method of reactive voltage partition with spectral clustering analysis based on principal eomponent analysis （PCA）. According to the operation and structural feature of the power grid,the method takes the weighted adjacency matrix of the network topology as a simplified topology model of the reactive partition. The principal component of the partitioning model is analyzed with the PCA, and the number of the principal components is taken as the dimension standard of selecting the feature vector in spectral clustering, therefore the division of the power grids is done hy the improved method of the spectral clustering algorithm. At the same time, the new modularity of the indicators is established by combining the deficiency of reactive power and the modularity of Q function, ensuring the rationality of the partition numbers and improving regional reactive power equilibrium. The reactive power reserve of the partitioning results is checked and adjusted to get the final partitioning scheme. Finally, the validity of this partitioning method is verified by the simulation and test of standard IEEE30-node.

关 键 词：复杂网络理论 PCA 谱聚类 无功储备度 模块度Q函数 

分 类 号：TM732[电气工程—电力系统及自动化]