摘要
提出一种基于事件检测的住宅用电负荷非侵入式识别方法。改进滑动窗双边累计和事件检测方法中确定事件发生阈值的设定措施,并在程序能够检测事件发生时刻的基础上,进一步提出事件发生后进入稳态时刻的检测方法。在负荷识别问题中,提出采用C4.5决策树算法和最近邻法结合的分类分层负荷识别算法。经过试验验证,事件检测算法和负荷识别算法均取得良好的负荷识别效果。
A event-based non-intrusive load monitoring method of residential building is proposed. The measures for setting event occurrence threshold in bilateral CUSUM event detection method of sliding window are improved and the detection methods in steady state after the occurrence of event are further put forward on the basis of the program able to detect the occurrence time of the event. In regard to the problem of load monitoring, a categorical and hierarchical load monitoring algorithm combining C4.5 decision tree algorithm with nearest neighbor algorithm is proposed. Through experimental verification, both the event detection algorithm and load monitoring algorithm have achieved good load monitoring effect.
出处
《建筑电气》
2017年第7期57-64,共8页
Building Electricity
关键词
非侵入式负荷识别
事件检测
滑动窗双边累计和算法
决策树
最近邻法
分类分层负荷识别
特征提取
负荷识别
Keydetection
window
categoricalwords： non-intrusive load monitoring
event bilateral CUSUM algorithm of slidingdecision tree
nearest neighbor algorithm
and hierarchical loadextraction
load monitoringmonitoring
feature
