摘要
为了提高电力企业管理水平,保证数据计量的准确性,提出了孤立森林算法下电力负荷异常数据辨识方法。利用拉格朗日插值方法插补电力负荷数据缺失值,标准化处理插补后的数据,获得标准化处理后的数据。根据随机解耦特征分解方法分解标准的电力负荷数据的谱特征,得到数据稀疏异质特征点集合。基于数据特征运用孤立森林算法划分电力负荷数据,构造iTree,运用异常分值完成异常数据辨识。通过实验证明所提方法能够精准辨识电力负荷异常数据,辅助相关人员及时作出修复决策,保证电力系统安全性。
In order to improve the management level of power enterprises and ensure the accuracy of data measurement,this paper put forward a method for identifying abnormal data of power load based on isolated forest algorithm.First,Lagrange interpolation method was adopted to interpolate the missing values of power load data,and then the interpolated data was standardized.Moreover,the spectral characteristics of the standardized power load data were decomposed by the random decoupling decomposition method,and then a set of sparse and heterogeneous feature points was obtained.Based on the data features,the isolated forest algorithm was used to segment the power load data and construct iTrees.Finally,the identification of abnormal data was completed by abnormal scores.The experimental results prove that the proposed method can accurately identify abnormal data of power load and assist relevant personnel in making repair decisions in a timely manner,thus ensuring the safety of power system.
作者
杨雪
陈巍
刘静
李昌利
YANG Xue;CHEN Wei;LIU Jing;LI Chang-li(Nanjing Institute of Technology,Jiangsu Nanjing 211167,China;Nanjing University of Information Science&Technology,Jiangsu Nanjing 210044,China)
出处
《计算机仿真》
2025年第3期140-144,共5页
Computer Simulation
基金
国家电网公司总部科技项目(5108-202218280A-2-296-XG)。
关键词
孤立森林算法
电力负荷
异常数据辨识
频谱特征
数据标准化处理
Isolated forest algorithm
Power load
Identification of abnormal data
Spectral characteristics
Data standardization
