基于电厂同步多点测量数据的相关聚类分析  

Correlation clustering analysis based on synchronous multi⁃point measurement data of power plants

作　　者：伍英伟 庞万禄 黄一晟 柯志冬 谢亚军 WU Yingwei;PANG Wanlu;HUANG Yisheng;KE Zhidong;XIE Yajun(Centralized Control Center(Hongshui River)of Guangxi Guiguan Eletric Power Co.,Ltd.,Nanning 530000,China)

机构地区：[1]广西桂冠电力股份有限公司(红水河)集控中心,广西南宁530000

出　　处：《电子设计工程》2024年第10期83-86,91,共5页Electronic Design Engineering

基　　金：基于桂冠电力生态云平台的智慧发电运行应用建设项目(AJ2022010)。

摘　　要：为了对电厂的同步多点测量数据进行分析,使用全局指标代替经典参数的方法来减少分析数据量,并检查相位值之间的相关性。利用k-means算法进行聚类分析,检测和排除数据集中的异常值,以确保从一般角度实现评估,得出电能质量点之间的关系,并对聚类分析方法清洗后的电厂数据进行多点测量的相关性分析。电能质量的不同测量点之间的相关性是使用Pearson系数和Spearman秩系数来实现的。分析结果表明,电压包络线和闪烁指示器之间的相关性降低,但仍然显著,全局指标之间存在较多的显著相关性,可以为解决电厂的电能质量问题提供参考。In order to analyze the synchronous multi⁃point measurement data of power plants,the method of using global indicators instead of classical parameters is used to reduce the amount of analysis data and check the correlation between phase values.Using the k⁃means algorithm for clustering analysis,detecting and excluding outliers in the dataset to ensure evaluation from a general perspective,obtaining the relationship between power quality points,and conducting correlation analysis of multi⁃point measurements on the cleaned power plant data using the clustering analysis method.The correlation between different measurement points of power quality is achieved using Pearson coefficient and Spearman rank coefficient.The analysis results show that the correlation between the voltage envelope and the flicker indicator decreases,but it is still significant.There is a significant correlation between the global indicators,which can provide reference for solving power quality problems in power plants.

关 键 词：相关分析 聚类分析 电能质量 多点测量 虚拟电厂 

分 类 号：TN87[电子电信—信息与通信工程]