Semi-Supervised Machine Learning for Fault Detection and Diagnosis of a Rooftop Unit

Most heating, ventilation, and air-conditioning (HVAC) systems operate with one or more faults that result in increased energy consumption could lead to system failure over time. Today, most building owners are performing reactive maintenance only may be less concerned able assess the health of until catastrophic occurs. This is mainly because do not previously have good tools detect diagnose these faults, determine their impact, act on findings. Commercially available fault detection diagnostics (FDD) been developed address this issue potential reduce equipment downtime, costs, improve occupant comfort reliability. However, many require an in-depth knowledge behavior thermodynamic principles interpret results. In paper, supervised semi-supervised machine learning (ML) approaches applied datasets collected from operating field develop new FDD methods help see value proposition proactive maintenance. The study data was packaged rooftop unit (RTU) HVAC running under normal conditions at industrial facility Connecticut. paper compares three different for classification a real-time RTU using learning, achieving accuracies as high 95.7% few-shot learning.