摘要
识别用户的负荷用电特性与用电行为是智能电网的重要研究内容之一。该文提出一种基于U-I轨迹曲线精细化识别的非侵入式负荷监测方法,实现对用户负荷有效的非侵入式监测。首先,利用拟合优度检验捕捉用电器投切事件,提取负荷的有功、无功功率变化量以及U-I轨迹3类特征。然后,进行两阶段负荷识别:第一阶段利用考虑初始优化的k-means算法对有功、无功变化量进行聚类,并压缩聚类个数,将功率特征相近的用电器聚为同组,得到一阶段识别判据,实现负荷粗辨识;第二阶段针对一阶段存在的识别盲区,构建卷积神经网络模型,以二维U-I轨迹图作为输入,通过卷积神经网络自动提取轨线的有效特征,实现一阶段盲区负荷的精细化识别。最后,利用BLUED数据集进行方法有效性的验证。
It is one of the important research contents of smart grid to analyze the user’s load characteristics and electricity consumption behavior.In this paper,a non-intrusive load monitoring method based on refined identification of U-I trajectory curve is proposed to realize the effective non-intrusive monitoring of the user’s load switching.Firstly,the goodness of fit test is used to capture the switching events of the electrical appliances,and the three kinds of characteristics of the active and reactive power variation and the U-I trajectory are extracted.Then,two stages of load identification are presented:in the first stage,the k-means algorithm considering initial optimization is used to cluster the active and reactive power changes,and the number of clusters is compressed.The electrical appliances with similar power characteristics are clustered into the same group,and the identification criteria of the first stage are obtained to realize the rough identification of load;In the second stage,the convolution neural network model is constructed to solve the blind areas in the first stage.The U-I trajectory is used as the input,and the effective features of the trajectory are extracted automatically by using the convolution neural network to realize the fine identification of the blind area load in the first stage.Finally,the BLUED dataset is used to verify the effectiveness of the method.
作者
汪颖
杨维
肖先勇
张姝
WANG Ying;YANG Wei;XIAO Xianyong;ZHANG Shu(College of Electrical Engineering,Sichuan University,Chengdu 610065,Sichuan Province,China)
出处
《电网技术》
EI
CSCD
北大核心
2021年第10期4104-4113,共10页
Power System Technology
基金
国家自然科学基金资助项目(52077145,52007126)。
