Kernel-Based Non-parametric Clustering for Load Profiling of Big Smart Meter Data

The emergence of smart meters has enabled the new energy efficiency services in an automatic fashion. With the information and communication technology, the smart meters are devised to gather and communicate the information of electricity suppliers and residential electricity consumers to ameliorate the efficiency of power distribution as well as the sustainability of the power resources. Due to the enormous amount of electricity consumers, the analysis of the big data produced by the smart meters is a crucial challenge faced by the electricity companies and researchers. In this paper, we analyze the big data based on the smart meter readings collected in the Houston area. The statistical properties of the data is investigated such that the behaviors of the consumers can be better understood. Moreover, the kernel PCA analysis and non-parametric clustering of the data gives a comprehensive guidance on what are the potential clusters of the customers and how to allocate the power more efficiently.