摘要
为了避免电流、电压波动的影响并准确可靠地获取负荷事件发生点和与其相关联的辨识特征,提出一种基于粒子重采样的负荷辨识特征最优选取方法。该方法首先根据滑动窗双边累积和(cumulative sum,CUSUM)负荷事件检测机理,通过类内方差最小规则获取负荷事件变点,定位负荷事件发生变点时刻。然后根据这个变点,采用粒子重采样方法获取变点前后稳定可靠的负荷信息,按照重要度原则选择最佳负荷特征信息。为了度量负荷辨识特征的有效性,引入模糊隶属度度量方式评价实验得到的负荷特征数据。真实测试结果表明,该方法提取的负荷能够较好地与数据库中的特征进行匹配,为后续基于数据库的准确负荷辨识奠定基础。
This paper proposes an optimal selection method of load identification features based on particle resampling to avoid the influence of current and voltage fluctuations,and to accurately and reliably obtain the identification characteristics associated with the occurrence of load events.In this method,it firstly adopts the sliding window double cumulative sum(CUSUM)load event detection mechanism.It combines the minimum inner-class variance rule to obtain the load event change point and locate the load event occurrence time.Then according to the change point,the particle resampling method is used to obtain the stable and reliable load information before and after the change point,and the optimal load characteristic information is selected based on the importance principle.Furthermore,in order to measure the efficiency of the load identification feature,the fuzzy membership degree measurement method is introduced to evaluate the obtained feature data.The real experimental results show that the load extracted by the method can match the features in the database well,and the research lays a foundation for subsequent accurate database-based load identification.
作者
林小红
周东国
肖勇
李秋硕
何恒靖
胡文山
LIN Xiaohong;ZHOU Dongguo;XIAO Yong;LI Qiushuo;HE Hengjing;HU Wenshan(School of Electrical Engineering and Automation,Wuhan University,Wuhan 430072,China;China South Power Grid Research Institute,Guangzhou 510663,China)
出处
《武汉大学学报（工学版）》
CAS
CSCD
北大核心
2020年第10期894-900,907,共8页
Engineering Journal of Wuhan University
基金
中国南方电网有限责任公司科技项目(编号:ZBKJXM20170079)。
关键词
粒子重采样
特征提取
非侵入式
负荷辨识
particle resampling
feature extraction
non-intrusive
load identification
