Energy Outlier Detection in Smart Environments

Despite a dramatic growth of power consumption in households, less attention has been paid to monitoring, analyzing and predicting energy usage. In this paper, we propose a framework to mine raw energy data by transforming time series energy data into a symbol sequence, and then extend a suffix tree data structure as an efficient representation to analyze global structural patterns. Then, we use a clustering algorithm to detect energy pattern outliers which are far from their cluster centroids. To validate our approach, we use real power data collected from a smart apartment testbed during two months.