Before the Breakdown: Using Predictive Modeling to Predict HVAC Maintenance

This project investigates how computer modeling can predict HVAC maintenance problems before they cause a major breakdown. HVAC systems can become less efficient or fail when parts become dirty or worn, and problems may not be noticed until the system stops working. The research question is: Can a computer model use HVAC sensor data to predict when maintenance will be needed before a breakdown occurs?

If a computer model monitors temperature, airflow, and energy use, it may predict maintenance problems earlier because unusual changes can show that the system is not working normally. The independent variables will be HVAC conditions and sensor measurements, while the dependent variable will be whether the model predicts that maintenance is needed. Other factors include filter condition, outdoor temperature, system age, and amount of use.

The project will research how HVAC systems work, common maintenance problems, sensors, predictive maintenance, and how dirty filters and worn parts affect performance. The model will use data such as indoor and outdoor temperature, airflow, energy use, operating time, filter condition, and maintenance events. The procedure will include collecting data, organizing it, identifying normal and unusual patterns, creating and testing the model, comparing predictions with known maintenance conditions, and calculating accuracy.

I expect the model to identify problems before major breakdowns. Limitations include limited data, different HVAC systems, inaccurate sensors, weather effects, and false predictions. This project could make HVAC maintenance cheaper, easier, and more reliable.