山东电力技术2025，Vol.52Issue(12)：61-69,9.DOI:10.20097/j.cnki.issn1007-9904.240308
适应新增电器的非侵入式负荷监测增量学习方法
Non-intrusive Load Monitoring Incremental Learning Method for New Appliances
摘要
Abstract
Fully mining the fine-grained energy consumption information of household appliances is helpful for power users to make wise energy-saving decisions.Non-intrusive load monitoring(NILM)is aimed at identifying the operating state of individual appliances based on measurements from a smart meter at the user's power entrance.At present,machine learning-based methods have shown good experimental results in load identification,but these results are usually based on the assumption that the type and number of loads are fixed.However,in real users,new load category data will continue to be generated over time.Therefore,the machine learning model should be able to continuously learn new load data,thereby enhancing its recognition ability.To solve this problem,a NILM incremental learning method adapted to new electrical appliances is proposed.The method only needs to store a small amount of prototype data of known electrical appliances and dynamically update it,which can continuously identify new electrical appliances without forgetting the learned knowledge of known electrical appliances.The effectiveness and superiority of the proposed model are confirmed by the comparative experimental results on the open dataset.关键词
非侵入式电器分类/增量学习/暂态电流特征/电器分类Key words
non-intrusive appliance classification/incremental learning/transient current characteristics/appliance classification分类
信息技术与安全科学引用本文复制引用
YANG Guochao,JIA Minghui,LI Zimo,BAI Xinyu,YU Puyao,LIU Yao..适应新增电器的非侵入式负荷监测增量学习方法[J].山东电力技术,2025,52(12):61-69,9.基金项目
国网天津市电力公司科技项目(城东-研发2023-03).Science and Technology Project of State Grid Tianjin Electric Power Company(Chengdong-R&D 2023-03). （城东-研发2023-03）
