Cluster analysis of university campus smart meter data

Recently, research on the field of energy efficiency of public buildings has drawn particular attention, following the current European energy policy. Hence, towards this direction is the study of the electricity consumption of buildings, since it can lead to significant information regarding their energy performance. Load profiling tools are required for the analysis of energy data, accessible through smart metering systems. The methodology proposed in this paper aims to analyze the case of Kimmeria Campus of Democritus University of Thrace, in Xanthi, Greece, through clustering. The K-means++ algorithm is considered, which is the most widely adopted algorithm. The methodology includes, pre-processing steps and transformation of real power data by means of the Fast Fourier Transform. The obtained representative load curves are related to monthly profiles calculated using statistical analysis.