基于L_1范数和隐马尔可夫链的用电设备非侵入辨识方法研究

Non-Intrusive Load Identification Methods Based on L_1-Norm and Hidden Markov Chain Model

针对用电设备非侵入式负荷辨识精度不高、效率低等问题,在综合隐马尔可夫链模型(Hidden Markov chain Mode)后,提出了一种基于L1范数的稳态特征非侵入负荷辨识的方法。该方法将稳态谐波电流及稳态有功功率作为特征参数建立特征数据库,再通过边沿检测获取准确的稳态状态段,然后运用L1范数最小化算法进行迭代计算,同时以训练完毕的隐马尔可夫链模型对其进行修正,有效解决了传统模式识别类算法稳定性差、辨识效果不佳的缺陷。同时通过仿真和实验分析,对所提出的方法和传统的辨识方法进行了对比分析,结果证明该方法具有良好的辨识精度和辨识效率。

This paper presents a non-intrusive load identification method to improve accuracy and efficiency of non-intrusive load monitoring by using L1 norm and the hidden Markov chain model.L1 norm is used to identify the basic type of equipments,then combined with the frequency phase angle of the electrical equipment and harmonic distortion rate,load comprehensive identified by hidden Markov chain model.