摘要
针对现有数据聚类方法难以对电力系统负荷数据进行有效聚类的问题,该文结合改进k-means算法,完成电力负荷数据聚类方法设计。该研究基于电力负荷数据中心点生成过程,构建中心点间距与类簇距离判定函数,筛选电力负荷数据聚类中心。确定聚类中心后,采用数据分离方法完成正常负荷数据和异常负荷数据的分离,在分离过程中应保证数据连续,以避免潜在有用数据丢失。利用改进的k-means算法分析电力负荷数据,计算不同种类数据间的欧氏距离。设定指针矩阵,融合不同类中心点,对原始数据区间规范化操作,获取不同簇的负荷数据聚类通道传输功率谱密度。将数据依次分配到不同簇上,实现电力负荷数据聚类。由实验结果可知,该方法站点1数据聚类范围为0.3~0.48 pu,站点2数据聚类范围为0.34~0.47 pu,优于对比方法,与理想聚类范围最贴近,具有良好的聚类效果。
Aiming at the problem that existing cluster analysis methods are difficult to cluster power system load data effectively,this research combined with improved k-means algorithm to complete the design of power load cluster analysis method.Based on the generation process of power load data center points,this study first constructs the center point spacing and cluster distance decision function,and screens power load cluster analysis centers.After determining the clustering center,the data separation method is used to separate normal load data and abnormal load data.During the separation process,data continuity should be ensured to avoid potential useful data loss.The improved k-means algorithm is used to analyze the power load data and calculate the Euclidean distance between different types of data.Set the pointer matrix,fuse different kinds of center points,standardize the original data interval,and obtain the spectral density of load cluster analysis channel transmission power of different clusters.The data is distributed to different clusters in turn to achieve power load cluster analysis.From the experimental results,it can be seen that the cluster analysis range of this method is 0.3~0.48 pu for Site 1 and 0.34~0.47 pu for Site 2,which is better than the comparison method and is the closest to the ideal clustering range,with good clustering effect.
作者
吕相沅
陈安琪
刘青
程昱舒
LV Xiangyuan;CHEN Anqi;LIU Qing;CHENG Yushu(Marketing Service Center of State Grid Shanxi Electric Power Company,Taiyuan 030000,China)
出处
《电子设计工程》
2024年第20期121-124,129,共5页
Electronic Design Engineering
基金
国网山西省电力公司科技项目资助(52051L20000A)。
作者简介
吕相沅(1991-),女,山西运城人,硕士研究生,工程师。研究方向:计量设备及辅助装置检定/检测、全性能试验、计量质量技术监督、用电负荷辨识等。
