摘要
居民用户作为智能电网的重要消耗端,合理用电对缓解能源危机起着至关重要的作用,用电量的分项计量及实时反馈能够引导用户自行优化用能习惯,同时帮助电网侧挖掘用户侧的节能潜力和需求响应潜力.非侵入式负荷监测是用电量分项计量的实现途径,本文针对现有高精度的基于深度学习的负荷识别算法运算复杂度高,无法用于家庭嵌入式设备的问题,提出利用无需训练过程的k最近邻(k-Nearest Neighbor,kNN)算法作为负荷识别模型.首先对标准kNN算法容易误判少数类的缺陷采用加权方式进行改进,然后针对V-I轨迹缺失数值特征的不足提出了基于综合相似度的类别判决方法,最后利用数据集和实验室数据验证了上述算法的有效性.
As an important consumption end of the smart grid,residential users play an important role in alleviating the energy crisis.The sub-metering and real-time feedback of electricity consumption can guide users to optimize their energy consumption habits and help the grid side to tap the energy-saving potential and demand response potential of the user side.Non-Intrusive Load Monitoring(NILM)is the premise of electricity consumption sub-metering.Aiming at the problem that the current high-precision load identification model based on deep learning has high computational complexity and cannot be applied to the home embedded devices,this paper proposes to use the k-Nearest Neighbor(kNN)algorithm without training process as the load identification model,to improve the defects of the traditional kNN algorithm,weight distribution method is used,and at the same time,aiming at the deficiency of missing numerical features of V-I trajectory,a class decision method based on comprehensive similarity is proposed,the validity of the above algorithm is verified by using public dataset and laboratory data.
作者
延菲
张瑞祥
孙耀杰
陶余会
黄国平
孙伟涛
YAN Fei;ZHANG Ruixiang;SUN Yaojie;TAO Yuhui;HUANG Guoping;SUN Weitao(School of Information Science and Engineering,Fudan University,Shanghai 200433,China;Institute for Six-sector Economy,Fudan University,Shanghai 200433;Shanghai Engineering Research Center for Artificial Intelligence Integrated Energy System,Shanghai 200433,China;Shanghai Fudan Forward Science and Technology Co.,Ltd,Shanghai 200433,China;CECEP Solar Energy Technology(Zhenjiang)Co.,Ltd,Zhenjiang,Jiangsu 212132,China;School of Electronic Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)
出处
《复旦学报（自然科学版）》
CAS
CSCD
北大核心
2021年第2期182-188,共7页
Journal of Fudan University：Natural Science
基金
国家重点研发计划(2018YFB1500904,2019YFB2103200)
2019年度上海市工程技术研究中心建设计划(19DZ2252000)
2020年第一批上海市信息化发展专项资金(智慧城市建设和大数据发展)(202001015)。
关键词
负荷识别
KNN算法
二值V-I轨迹
综合相似度
load identification
kNN algovithm
binary V-I trajectory
comprehensive similarity
