Novel Approach for Predictive Maintenance of Air Handling Units Using Machine Learning Algorithms

Traditional maintenance of Air Handling Units (AHUs) in buildings is often reactive or time-based, resulting in unexpected failures, downtime, and higher costs. This paper explores Machine Learning (ML) for predictive maintenance (PrM) of AHUs, using user-provided data. A dataset from Granderson and Lin (2019) demonstrates how ML models can classify AHU conditions as faulty or normal. This research discusses suitable ML algorithms for AHU PrM, including supervised learning, and highlight benefits such as reduced downtime, cost savings, improved efficiency, and enhanced occupant comfort. The methodology covers data acquisition, preprocessing, model selection, training, and evaluation.