Abstract:Solved the problem of data loss and noise in the process of data collection and transmission of smart electric appliance network, and improved the power data quality,which can be effectively use all kinds of big data analysis and prediction algorithms for the electric appliance cloud platform. Based on the summary of the characteristics of data collection and data transmission of the smart electric appliance network, and the analysis of the data quality requirements of the cloud platform, this paper proposes a method of power data preprocessing of the smart electric appliance network.After normalizing the electric appliance data collected by smart electric appliance terminals, normalization data are extracted from noise, fuzzy and random data by means of clustering algorithm, and the clustering effect of k-means clustering and density based spatial clustering of applications with noise (DBSCAN) algorithms is compared and verified. Compared with k-means clustering algorithm, DBSCAN algorithm can automatically obtain the clustering number of complex shape data sets while detecting data noise points, which is more suitable for electric appliance data preprocessing of the smart electric appliance network.