基于VMD样本熵与随机森林的非侵入式设备状态检测方法

Non⁃invasive equipment status detection method based on VMD sample entropy and random forest

非侵入式负荷监测可以保护用户隐私并对用电情况做出细分,为了提高设备状态检测辨识精度,提出一种基于变分模态分解样本熵特征数据集结合随机森林分类的非侵入式负荷状态检测方法。采用变分模态分解算法利用其良好的抗噪性、鲁棒性对暂态特征进行提取;计算模态样本熵,将处理后的数据集作为决策树、支持向量机和随机森林等算法的输入进行负荷状态辨识。实验分析表明,随机森林的强分类性使得负荷辨识结果较好,且拟合效果优秀,成功结合了所提算法并解决了单设备运行和多设备运行时的非侵入式设备状态的检测问题。

Non⁃intrusive load monitoring can protect user privacy and subdivide power consumption.A non⁃intrusive load state detection method based on variational mode decomposition(VMD)sample entropy feature data set combined with random forest classification is proposed to improve the accuracy of equipment state detection and identification.The variational mode decomposition(VMD)algorithm is used to extract the transient features on virtue of its good anti⁃noise characteristics and robustness,and then the modal sample entropy is calculated.The processed data sets are taken as the input of the decision tree algorithm,support vector machine(SVM)and random forest algorithm to carry out load state identification.Experimental analysis shows that the strong classification of the random forest makes the results of load identification better and the fitting effect excellent.It successfully combines the proposed algorithm and solves the problem of non⁃intrusive device status detection during single⁃equipment operation and multi⁃equipment operation.