Monitoring building HVAC runtime using low-cost acoustic sensors.

This paper evaluates whether low-cost acoustic sensors can be used to predict HVAC on/off state and runtimes by comparing acoustically inferred air handler unit (AHU) operation with power level measurements from a dedicated AHU circuit. We demonstrate and evaluate this approach in a full-scale laboratory home designed for energy performance testing. A threshold-based state inference algorithm applied independently to the power and noise data showed 96.8% agreement, suggesting that overall A-weighted sound level can be used to estimate AHU state for two-state systems with careful sensor placement.