基于负熵估计的居民用电负荷非侵入式分解算法  

A Non-Intrusive Decomposition Algorithm for Resident Power Load Based on Negative Entropy Estimation

作　　者：武昕[1] 王震[1] 

机构地区：[1]华北电力大学电气与电子工程学院,北京市昌平区102206

出　　处：《电网技术》2017年第3期931-937,共7页Power System Technology

基　　金：国家重点研发计划项目课题(2016YFB0901104)资助;中央高校基本科研业务费专项资金资助项目(2016MS13)~~

摘　　要：非侵入式负荷监测是实现能效跟踪与智能用电的重要技术,其中,负荷辨识方法是非常重要的内容。为此研究了一种非侵入监测机制下的居民负荷快速分解算法。利用非侵入负荷监测模式下负荷总电流信号是负荷单独运行时电流信号混合叠加的特点,将负荷分解问题有效建模为盲源分离问题。对混合电流信号进行白化处理,并构建解混矩阵,基于负熵最大化判别准则形成了有效的居民用电负荷分解算法,并构建评价函数对分解结果进行量化分析。为了验证算法的有效性,利用实际采集的用电数据进行负荷分解,均能够准确地从混合电流信号中分解出各个单独的叠加信号,即得到当前参与运行的用电负荷,并能够根据相似系数确定负荷类型,实现负荷辨识。Non-intrusive load monitoring is important to energy efficiency tracking and intelligent power. For non-intrusive load monitoring, load identification is of great importance. Therefore, a fast decomposition method of resident power load under non-intrusive monitoring mechanism is studied in this paper. Based on characteristics of total load current signal mixed with load current signal of separate runtime, load decomposition problem is modeled as a blind source separation problem effectively. Firstly, the mixed current signal is whitened. Secondly, de-mixing matrix is constructed. At last, resident power decomposition algorithm is formed based on negative entropy maximization criterion. An evaluation function is constructed to quantify decomposition result. Real sampling load data are used to verify effectiveness of the algorithm. Each current signal can be separated from mixed current signals accurately, that is, each power load of runtime can be obtained. According to similarity coefficient, load types can be determined to realize load identification.

关 键 词：非侵入式负荷监测 负荷分解 居民负荷 负熵估计 盲源分离 

分 类 号：TM721[电气工程—电力系统及自动化]