Hybrid Noninvasive Load Identification with Combining PCA-DBSCAN Clustering and Comprehensive Correlation Analysis

To improve the accuracy of power load monitoring, A hybrid nonintrusive load identification method based on density-based spatial clustering of applications with noise with principal component analysis (abbr. PCA- DBSCAN) was proposed. Firstly, in allusion to the problem of high dimensionality of the original load features, the PCA algorithm was used to reduce the dimensionality of the original feature data to construct a load feature template library, simultaneously, the load current waveform was obtained to construct the load current template library. Secondly, the DBSCAN clustering algorithm was used to unsupervised cluster the samples in the load feature template library to extract the centers of each cluster. Thirdly, the Euclidean distance between the load to be identified and the clustering centers of each feature template library was calculated for load feature matching; and the comprehensive correlation degree between the current waveform of the load to be identified and each of that in the current waveform template library was calculated for realizing load current waveform matching. Finally, by mixing the two matching results, load identification was taken synthetically for realizing high reliable identification. The simulation results of a testing dataset with electricity consumption data show that the indexes of the proposed method are over 96%.