基于量子遗传和核模糊聚类的低压台区户变关系识别

姚力; 章江铭; 倪琳娜

姚力,章江铭,倪琳娜.基于量子遗传和核模糊聚类的低压台区户变关系识别[J].电测与仪表,2020,57(20):106-113.
Yao Li,Zhang Jiangming,Ni Linna.User-Transformer relation identification based on quantum genetic and kernel fuzzy clustering[J].Electrical Measurement & Instrumentation,2020,57(20):106-113.

基于量子遗传和核模糊聚类的低压台区户变关系识别

User-Transformer relation identification based on quantum genetic and kernel fuzzy clustering

DOI：10.19753/j.issn1001-1390.2020.20.015

中文关键词: 户变关系识别过零偏移核模糊聚类量子遗传小生境协同进化动态调整策略 Hadamard门变异策略

英文关键词:User-transformer relation identification Zero-crossing shift Kernel fuzzy clustering Quantum genetic Niche coevolution strategy Dynamic adjustment strategy Hadamard gate mutation strategy

基金项目:国家电网总部科技项目（基于跨平台多源数据融合的复杂低压台区拓扑识别及线损计算应用关键技术研究，5600-201919168A-0-0-00）

作者中文名	作者英文名	单位
姚力	Yao Li	国网浙江省电力有限公司营销服务中心
章江铭	Zhang Jiangming	国网浙江省电力有限公司营销服务中心
倪琳娜	Ni Linna	国网浙江省电力有限公司营销服务中心

摘要点击次数: 1478

中文摘要:

户变关系对于营配融合、线损分析等业务的开展具有重要作用。为了得到准确的户变关系,本文提出一种基于量子遗传和模糊聚类的户变关系识别方法。主要思想为：由于不同台区电能表的电压过零偏移特征不同,采用核模糊C均值聚类对电能表电压的过零偏移进行分类,通过与变压器端的过零偏移比较,实现户变关系识别。首先,为了提高聚类精度和效率,采用量子遗传算法对模糊C均值聚类的聚类中心和核参数进行优化,并提出一种基于类间距离和类内距离的适应度函数构造方法。另外,为了解决遗传算法易陷入局部极值的问题,分别引入小生境协同进化策略、动态调整策略、Hadamard门变异策略,提高算法寻优能力。其次,通过对benchmark函数和UCI数据集特征的仿真测试,验证了本文提出方法比标准核模糊C均值聚类具有更高的聚类精度和运算效率。最后,采用本文方法对实际的台区变压器和电能表数据进行归属识别,结果表明,本文提出方法得到的结果与真实户变关系一致,具有较好的应用效果。

英文摘要:

User-transformer relation is very important to the service deployment of business distribution integration, line loss analysis, etc. In order to obtain the accurate affiliation relation, a new user-transformer relation identification method based on quantum genetic algorithm (QGA) and kernel fuzzy C-means clustering (KFCM) is proposed in this paper. The main idea is: since the power meter of different areas has different voltage zero-crossing shift features, the affiliation identification is realized through comparing the zero-crossing shift near transformer with the classification result of zero-crossing shift on power meter which is obtained by KFCM. At first, in order to improving the accuracy and efficiency of clustering, the QGA is introduced to optimize the cluster center and kernel parameters of KFCM and a fitness function construction method based on inner-class distance and between-class distance is present. Besides, the niche coevolution strategy, the dynamic adjustment strategy and the Hadamard gate mutation strategy are proposed to enhance the optimization ability of QGA. Then, the simulation tests on benchmark function and UCI dataset character verify that the proposed method in this paper has better cluster accuracy and efficiency than standard KFCM. At last, the proposed method is applied to solve the user-transformer relation identification of practical low voltage area. Result shows that the power meter affiliation obtained by proposed method is the same as the real affiliation. The application effect is very good.

查看全文查看/发表评论下载PDF阅读器

关闭