| 引用本文: | 徐伟枫,华锦修,余涛,刘前进,蓝超凡.计及电器状态关联规则的非侵入式负荷分解[J].电力自动化设备,2020,40(4): |
| XU Weifeng,HUA Jinxiu,YU Tao,LIU Qianjin,LAN Chaofan.Non-intrusive load decomposition considering association rules of appliances’ state[J].Electric Power Automation Equipment,2020,40(4): |
|
| 摘要: |
| 非侵入式负荷监测与分解(NILMD)是获取电器用电信息的关键技术，针对当前NILMD缺乏考虑不同电器关联运行的用电模式和电器状态的强波动性以致分解精度低的问题，提出一种计及电器状态关联规则的新型负荷分解方法。通过仿射传播聚类提取电器的运行状态，基于互信息熵，运用关联规则算法挖掘电器状态的关联性；调整含关联规则的样本权值并结合k近邻算法实现状态辨识；利用极大似然估计完成负荷功率分解。测试算例验证了所提方法的有效性和准确性。 |
| 关键词: 非侵入式负荷监测 仿射传播 互信息熵 关联规则 k近邻 极大似然估计 |
| DOI：10.16081/j.epae.202003015 |
| 分类号:TM714 |
| 基金项目:国家电网有限公司总部科技项目(能源互联网环境下的多源互联配电网及多样化用电方式的需求策略系统研究) |
|
| Non-intrusive load decomposition considering association rules of appliances’ state |
|
XU Weifeng, HUA Jinxiu, YU Tao, LIU Qianjin, LAN Chaofan
|
|
School of Electric Power, South China University of Technology, Guangzhou 510641, China
|
| Abstract: |
| NILMD(Non-Intrusive Load Monitoring and Decomposition) is the key technology to obtain electricity consumption information of appliances. Aiming at the problem of low decomposition accuracy which is caused because that current NILMD lacks of concerning about the electricity consumption pattern of associated operation between different appliances and high volatility of appliances’ state, a novel load decomposition method considering the association rules of appliances’ state is proposed. The operation state of each appliance is extracted by affinity propagation clustering. Based on mutual information entropy, the association rule algorithm is used to mine the association of appliances’ state. The sample weights with association rules are adjusted and combined with kNN(k-Nearest Neighbor) algorithm to realize the state identification. The maximum likelihood estimation is used to decompose load power. Test examples verify the effectiveness and accuracy of the proposed method. |
| Key words: non-intrusive load monitoring affinity propagation mutual information entropy association rule kNN maximum likelihood estimation |
