基于聚类的阶段理论线损快速计算与分析

发布时间：2018-05-24 11:51

本文选题：最近邻聚类 + 阶段理论线损　；参考：《电工技术学报》2015年12期

【摘要】：针对自动采集数据缺失及单一断面网损无法刻画电网长期运行的情况,在仅考虑节点注入功率与网络参数变化的基础上,提出一种基于断面聚类的阶段理论线损快速计算与分析方法。本文首先对断面网损的构成进行了分析;其次针对断面数据采集情况提取断面特征向量,并且在数据采集不完整时,提出一种基于粗糙集理论的核特征向量求取方法;通过量化节点注入功率与网络参数单位变化引起的网损增量计算出与特征向量匹配的聚类中心断面的权系数向量;最后利用改进的最近邻聚类对断面进行聚类分析,快速得出任意时刻理论线损值和阶段理论线损值。Matlab仿真结果验证了此方法的有效性。
[Abstract]:In view of the absence of automatic acquisition data and the inability of single section network loss to describe the long-term operation of the power grid, only the changes of node injection power and network parameters are considered. A fast calculation and analysis method of phase theoretical line loss based on cross section clustering is proposed. In this paper, the structure of cross-section network loss is analyzed firstly, secondly, the feature vector of cross-section is extracted according to the data acquisition of cross-section, and when the data acquisition is incomplete, a method of calculating kernel eigenvector based on rough set theory is proposed. By quantifying the increment of network loss caused by the change of node injection power and network parameter unit, the weight coefficient vector of the cluster center section matched with the eigenvector is calculated. Finally, the improved nearest neighbor clustering is used to cluster the section. The theoretical line loss value at any time and the theoretical line loss value at the stage are obtained quickly. The simulation results with Matlab verify the effectiveness of this method.
【作者单位】：燕山大学电力电子节能与传动控制河北省重点实验室;国网冀北电力有限公司青龙县供电分公司;国网黑龙江省电力有限公司;
【基金】：国家自然科学基金(61473246) 河北省自然科学基金(E2015203294) 中国博士后科学基金(2014M551049)资助项目
【分类号】：TM744

【参考文献】