Random Forests Model for HVAC System Fault Detection in Hotel Buildings

Heating, ventilation, and air conditioning (HVAC) systems are essential for maintaining a comfortable indoor environment in modern buildings. However, HVAC systems are known to consume a lot of energy, which can account for up to 50% of a building's energy consumption. Therefore, it is important to detect and troubleshoot problems in HVAC systems timely. Fault detection and diagnosis (FDD) techniques can help with HVAC monitoring and optimizing system performance for efficient use of energy. In this paper, we demonstrate how to create efficient fault detectors using physics-based modeling and machine learning. We show how to build a simulation model of a hotel building, which we then use to sample augmented data with typical faults commonly found in HVAC systems. We train predictive models using random forests (RFs). The results suggest that RFs can be used as stand-alone detectors for FDD, albeit their performance depends heavily on the data quality.