HVAC System Air Filter Maintenance: A Fuzzy Machine-Learning-Based System

ABSTRACT- Heating, ventilating, and air conditioner (HV AC) systems are crucial for the comfort of occupants in many industrial, commercial, residential and institutional buildings. There are many reasone to monitor the status of components of an HV AC system, such as to reduce maintenance costs and energy usage. Air filters are critical components that contribute to such reduction. However, in many cases, it is challenging to determine the optimal time to replace/clean a group of filters to save operational costs. This is because the air filter condition is difficult to observe as it is in the supply duct with limited space to install ideal sensory tools such as airflow sensors. This paper develops a fuzzy machine-learning-based system that relies on weak sensors that fit into the limited space. To implement the system, a differential air pressure sensor with two small tubes to measure the airflow gap before and after the air filter, and temperature sensors to monitor air quality, are installed in a real HV AC system. Results show that the fuzzy-logic-based machine learning approach achieves 98.82 % accuracy. The dashboard and early warning system provide a reliable system for smart maintenance, monitoring the live sensors' status, sending alarms, and timely air filter replacement.