基于特征加权KNN的非侵入式负荷识别方法

Non-intrusive load identification method based on feature weighted KNN

针对不同稳态特征对识别结果的影响程度不同,并考虑到不平衡数据集造成的少数类误判的问题,提出一种基于特征加权KNN的非侵入式负荷识别方法。首先,采用熵权法计算特征权重,利用特征权重改进特征距离的计算。其次,根据样本数量和对应算法k值计算得到表决权重,带入投票表决过程中,以此来增加少数类的分类准确性。实验结果表明,针对实测负荷数据集时,本文算法的平均识别准确率为93.4%,与KNN算法相比提高了2.8%;针对公开数据集时,本文算法的平均准确率和F1得分分别为86.8%和81.6%,要优于其他4种分类算法。

In view of the different influence of different steady-state features on the identification results, and considering the misjudgment of minority classes caused by unbalanced data sets, a non-invasive load identification method based on feature weighted KNN is proposed. Firstly, the feature weight is calculated by entropy weight method, and it is used to improved feature distance calculation. Secondly, the voting weight is calculated according to the number of samples and the k value of algorithm, which is brought into the voting process to increase the classification accuracy of minority classes. The experimental results show that the average recognition accuracy of algorithm in this paper is 93.4%, which is 2.8% higher than that of KNN algorithm. For public data sets, the average accuracy and F1 score of algorithm in this paper are 86.8% and 81.6%, which are better than the other four classification algorithms.