Experimental validation of machine learning- based predictive control algorithms for HVAC systems
Loading...
URL
Journal Title
Journal ISSN
Volume Title
School of Engineering |
Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Authors
Date
2025-01-01
Department
Major/Subject
Sustainable Energy in Buildings and Built Environment
Mcode
Degree programme
Master's Programme in Advanced Energy Solutions
Language
en
Pages
60
Series
Abstract
The significance of HVAC systems for buildings is immense as they consume a high portion of energy. This calls for the search optimized and energy-efficient solutions for the HVAC systems for a reduction in energy demand. Integration of Federated learning techniques with a machine learning (ML) predictive model is considered a novel solution that not only provides innovative solutions but also has the potential to improve the operational performance of HVAC systems. The experimental setup location of this thesis is at a university building at Aalto University. Three air handling units (AHUs) were selected to be considered as a part of the research. The values for building parameters and operational setpoints were shared using a building management system (BMS) network. This data was used to train the ML model on local datasets, creating an empirical graph network based on similarities between nodes to predict the supply air temperature parameter for adjusting the operational control of the AHUs. This will affect the overall energy requirement of the HVAC system of the building. The integration of an ML predictive model with real operative HVAC systems can provide us with real-time data, which can be communicated and monitored using a BACnet protocol network control (YABE) internet explorer was here used to access the BACnet protocol server through IT controllers. An IDA ICE simulation model was then developed and validated by using similar data for training the model to analyze the effect of predictive model control after integration with the HVAC system and perform an according energy use analysis.Description
Supervisor
Ferrantelli, AndreaThesis advisor
Eik, MarikaKeywords
HVAC system, air handling unit, IDA ICE simulation model, machine learning, predictive control, federated learning