Energy Performance Based Anomaly Detection in Non-Residential Buildings Using Symbolic Aggregate Approximation

Abstract

Building system faults in commercial and office buildings can result in a reduced occupant comfort and increased utility bills. Energy performance-based anomaly detection helps operators efficiently identify faults. In this work, a data-driven method for anomaly detection is presented. Using a symbolic aggregate method, the weekly energy demand profiles are statistically quantised and labeled to determine normal and abnormal building behaviours. A case study with three federal office buildings has been conducted to demonstrate the proposed method. The resulting technology provides building operators with easily-interpreted and actionable information for optimised building performance.