基于差量特征提取与模糊聚类的非侵入式负荷监测方法

发布时间：2018-04-06 00:19

本文选题：非侵入式负荷监测　切入点：电器特征分析　出处：《电力系统自动化》2017年04期

【摘要】：现有非侵入式负荷监测(NILM)方法主要将电器功率大小作为特征值,对于低功率电器识别的准确性不够,无法满足精细化智能用电的应用需求。文中分析了多种家用电器的功率和谐波特征,并选取低功率电器差异最大的频域谐波幅值作为新的特征。在此基础上提出一种新的NILM方法,该方法采用差量特征提取方法获取任意时刻的特征值变化量并引入信息熵的方法,通过计算簇间熵来确定最佳聚类数和负荷相似度;再通过模糊聚类实现电器负荷数量及种类的聚类识别。实验结果表明,文中提出的NILM方法在不同场景下均具有良好的可靠性和鲁棒性,采用谐波特征后识别准确性有明显提升。
[Abstract]:The existing non-invasive load monitoring NILM methods mainly take the power of the electrical apparatus as the eigenvalue, and the accuracy of the identification of low-power electrical appliances is not enough to meet the application requirements of refined and intelligent power consumption.In this paper, the power and harmonic characteristics of various household appliances are analyzed, and the harmonic amplitude in frequency domain, which is the largest difference between low power appliances, is selected as the new feature.On this basis, a new NILM method is proposed, in which the difference feature extraction method is used to obtain the variation of eigenvalues at any time and the information entropy is introduced to determine the optimal clustering number and load similarity by calculating the entropy among clusters.Then the number and category of electric appliance load are identified by fuzzy clustering.The experimental results show that the proposed NILM method has good reliability and robustness in different scenarios.
【作者单位】：华北电力大学电气与电子工程学院;
【基金】：国家高技术研究发展计划(863计划)资助项目(SS2015AA050203) 国家电网公司科技项目“智能电网用户行为理论与互动化模式研究” 中央高校基本科研业务费专项资金资助项目(2015XS05)~~
【分类号】：TM76

【参考文献】