基于FBE算法的非侵入式事件检测方法

Non-intrusive Event Detection Method Based on FBE Algorithm

鉴于基于非侵入式负荷监测(NILM)的智能电表将成为住宅和建筑物的重要组成部分,提出一种应用非侵入式负荷监测环境的滤波器式经验模态分解算法,即将采集到的总电压电流信号自适应地分解成各种窄带信号,从分解的这些窄带信号中提取有关设备开关事件的有用信息,以有效提取模态特征、检测开关事件,克服传统经验模态分解方法对负类事件分类准确度的缺陷,增加了整体负载分类的准确性;最后,在BLUED数据集上进行测试,相比传统算法精度80.56%,所提方法的识别精度为97.39%,验证了算法的实用性和优越性。

Smart meters based non-intrusive load monitoring(NILM)can be an important part of houses and buildings.This paper proposes an empirical mode decomposition signal analysis algorithm for non-intrusive load monitoring environment,which can adaptively decompose the collected total voltage and current signals into various narrowband signals,and extract the relevant information from these narrowband signals.The useful information of each device switching event can achieve the advantages of effectively extracting modal features and detecting switching events,overcoming the defects of classification accuracy of negative events in the traditional empirical mode decomposition method,and increasing the accuracy of overall load classification.Finally,the practicality and superiority of the proposed algorithm are verified by testing the algorithm and the empirical mode decomposition method for detecting events on the BLUED data set.The identification accuracy of the proposed method is 97.39%while the accuracy of the traditional method is only 80.56%.