微型电脑应用2025，Vol.41Issue(4)：93-97,5.
基于改进聚类算法的电力工程数据识别与标记方法研究
Research on Data Identification and Labeling Method for Electric Power Engineering Based on Improved Clustering Algorithm
摘要
Abstract
In view of the shortcomings of subjective judgment algorithm in identifying abnormal data of electric power engineer-ing,such as low accuracy and low efficiency,this paper proposes a method for identifying and marking related data based on improved clustering algorithm.This method is based on the density core clustering algorithm.For the problems of strong pa-rameter dependence and fixed clustering radius in the clustering process,the vector resultant force method is used to select the appropriate cluster center according to the density of data points.At the same time,the natural neighbor algorithm is used to adaptively select the data in the cluster,and then the model input is used as the data set.In the experimental test,the accura-cy,F1 value and normalized mutual information(NMI)index of the proposed algorithm can reach 0.77,0.75 and 0.73,re-spectively,which is generally ahead of other comparison algorithms in the artificial high-dimensional data set,and the test con-ducted in the real data set can also effectively find outliers.关键词
密度核心聚类/向量合力法/自然邻居算法/离群点检测/电力工程数据Key words
density core clustering/vector resultant force method/natural neighbor algorithm/outlier detection/electric power engineering data分类
信息技术与安全科学引用本文复制引用
刘士李,赵迎迎,陈付雷,沈磊,施晓敏..基于改进聚类算法的电力工程数据识别与标记方法研究[J].微型电脑应用,2025,41(4):93-97,5.基金项目
2022年国网安徽经研院科技类课题(B1120922000A) （B1120922000A）
