ORIGINAL RESEARCH article
Front. Built Environ.
Sec. Indoor Environment
This article is part of the Research TopicSmart Systems for Energy Efficiency and Indoor Environmental Quality in BuildingsView all articles
Integration of Digital Twins and Machine Learning for Predictive Maintenance Using APAR Method Rules in Non-residential Buildings
Provisionally accepted
- OsloMet - storbyuniversitetet Fakultet for teknologi kunst og design, Oslo, Norway
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
ABSTRACT Non-residential buildings are the largest global energy consumers, making Heating, Ventilation, and Air Conditioning (HVAC) system efficiency a critical area of focus. Within these systems, Air Handling Units (AHUs), as central components of HVAC systems, play a key role in regulating indoor climate, but are particularly prone to operational faults due to complex control dynamics. Addressing limitations in current fault detection methods, namely the lack of interpretability in Machine Learning (ML) models and the rigidity of rule-based systems, this study aims to develop a hybrid Predictive Maintenance (PdM) framework that is both transparent and scalable. The approach combines interpretable fault detection using the Air-Handling Unit Performance Assessment Rules (APAR), adaptive fault classification and prediction through machine learning (ML), and real-time monitoring via Digital Twin (DT) interface to enhance operational reliability and energy efficiency. To evaluate feasibility, the framework was deployed on an AHU in a non-residential facility in Grimstad, Norway, using six months of operational data with over 51,000 logged records. Evaluation results show that the hybrid approach significantly improves fault detection performance across both frequent and rare classes, with strong F1-scores and high recall for critical fault conditions. The DT component, integrated via pyRevit and a web-based dashboard, enables real-time fault visualization and supports maintenance planning. The findings validate that combining expert-driven rules with ML and DT technology provides a practical, accurate, and scalable solution for PdM in AHUs. This framework supports the transition from reactive to intelligent operations in building environments.
Keywords: predictive maintenance (PdM), Air Handling Units (AHUs), Fault detection and diagnosis (FDD), Air-Handling Unit Performance Assessment Rules (APAR), Machine Learning (ML), Digital twin (DT)
Received: 29 Oct 2025; Accepted: 16 Dec 2025.
Copyright: © 2025 Zabadi, Zemedkun, Hosamo and Kraniotis. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Haidar Hosamo
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.