基于改进kNN算法的非侵入式电力负荷监测仿真  

Simulation of Non-Invasive Power Load Monitoring Based on Improved kNN Algorithm

作　　者：黎捷 杨林峰[1] LI Jie;YANG Lin-feng(College of Computer Science,Guangxi University,Nanning Guangxi 530004,China)

机构地区：[1]广西大学计算机学院,广西南宁530004

出　　处：《计算机仿真》2025年第1期81-84,91,共5页Computer Simulation

基　　金：电力科技项目《能源数据中心城市大脑实施与建设》(2021-YF-TQSJ-02)。

摘　　要：负荷监测是保证电力系统正常使用过程中不可缺少的环节,但监测过程易受冗余负荷、电力复杂环境、强负载等问题的干扰。为了解决上述问题,提出基于改进kNN算法的非侵入式电力负荷监测算法。提取电力设备的基波电流、谐波电流与有功功率等特征,采用Fisher主元分析法筛选出有效特征,以提高非侵入式电力负荷监测精度。采用改进kNN算法计算特征的投影值,通过投影值与阈值的对比完成非侵入式电力负荷的监测。仿真结果表明,所提算法的监测耗时短、功率监测效果好、监测精度高。Load monitoring is an indispensable link in ensuring the normal use of the power system,but the monitoring process is susceptible to interference from redundant load,complex power environment,strong load,and other issues.To address these issues,a non-intrusive load monitoring algorithm based on the improved kNN algorithm was presented.Firstly,the characteristics of fundamental current,harmonic current,and active power of power equipment,were extracted.Then,Fisher principal component analysis was adopted to screen out effective features,thus improving the monitoring accuracy of non-intrusive power load.Moreover,the improved kNN algorithm was used to calculate the projection value of the feature.Finally,non-intrusive power load monitoring was completed by comparing the projec--tion value with the threshold.Simulation results prove that the proposed algorithm has short monitoring time,good power monitoring effect,and high monitoring accuracy.

关 键 词：负荷稳态特征参数 特征筛选 主元分析法 风险控制阈值 电力负荷映射 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]