基于DTW算法与稳态电流波形的非侵入式负荷辨识方法  

Non-intrusive Load Identification Method Based on DTW Algorithm and Steady-state Current Waveform

作　　者：祁兵[1] 董超 武昕[1] 崔高颖 

机构地区：[1]华北电力大学电气与电子工程学院,北京市102206 [2]国网江苏省电力公司电力科学研究院,江苏省南京市211103

出　　处：《电力系统自动化》2018年第3期70-76,共7页Automation of Electric Power Systems

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

摘　　要：非侵入家用负荷识别技术可以指导家庭用户合理安排用电,提高用能效率,同时也为电力部门提供家庭用电的数据支持,有利于了解负荷用电规律及趋势,完善电力规划。由于家庭用户用电负荷的稳态特征值存在相似和无规律的问题,现有的方法多采用高级算法对所有的用电负荷组合进行训练。针对现有采用负荷稳态特征值方法进行识别所存在的不足,考虑到家用负荷稳态波形具有独特性和叠加性,提出了一种利用动态时间弯曲(DTW)算法计算与模版库波形的距离来识别家用负荷的辨识方法。首先,建立负荷稳态波形模版库;然后,在电压满足一定条件的情况下,测量家庭用户电流的稳态波形;最后,利用DTW算法计算出最小距离进行识别。Non intrusive household load identification technology can guide reasonable arrangement of household users, improve energy efficiency, and provide domestic electrie data support to the power sector, which is conducive to understand the load electricity law and trends for improving power planning. As steady-state characteristic values of electric load for home users are similar and irregular, most existing methods need to use advanced algorithms to train all the combinations of electrical load. Faced with the existing problems with the steady load characteristic value method and considering the unique characteristics and the superposition of steady-state waveform of domestic load, a method based on the dynamic time warping （DTW） algorithm and template library waveform is proposed to identify the household load. First of all, a template library of steady-state waveform is established. Then, under certain voltage conditions, the current steady state waveform is measured. Finally, the DTW algorithm is used to calculate the minimum distance for identification.

关 键 词：非侵入式负荷辨识 负荷电流稳态波形 动态时间弯曲算法 电流稳态波形库 

分 类 号：TM714[电气工程—电力系统及自动化]