• Login
• Admin

Wei, Shuangyu (2023) A novel vision-based approach for accurate equipment usage detection for energy efficiency and indoor comfort level improvement in office buildings. PhD thesis, University of Nottingham.

PDF (Thesis correction) (Thesis - as examined) - Repository staff only until 1 January 2025. Subsequently available to Anyone - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (8MB)

Abstract

The aim of this study was to develop a real-time indoor equipment usage detection approach using computer vision and deep learning techniques to aid the adjustment of heating, ventilation, and air conditioning (HVAC) operations based on the actual demands in office and kitchen spaces. This could not only address the issue of under- or over-estimation of the building energy consumption but also maintain a comfortable indoor environment for occupants by adjusting HVAC’s operations based on the intuitive and real-time data.

The indoor equipment usage detection model was developed and implemented using Python and TensorFlow API. This work used AI-enabled cameras as the indoor detection technique and locally running trained deep learning algorithms to analyse and take action based on how equipment was utilised in the spaces. Experimental tests were carried out in the case study office and kitchen to assess the detection performance of the developed approach. The results indicate that the developed deep learning detection approach could achieve 82.52% accuracy in detecting office equipment and 91.42% in detecting kitchen equipment.

This work also compared the building energy performance of the developed approach with a conventional approach such as the use of static heating, cooling, and ventilation operation profiles through building energy simulation (BES) and on-site environmental measurement. The equipment usage profiles generated by the collected information from deep learning detection approach was fed into building energy models to evaluate the impact of using this approach in buildings. Simulation and environmental measurement results highlighted that following the profiles generated by deep learning detection techniques could make the HVAC system adapt to the actual demands to maintain a better indoor environment and essentially minimise the energy wastes arising when the supply of the HVAC systems is more than the demand.

Item Type:	Thesis (University of Nottingham only) (PhD)
Supervisors:	Calautit, John Boukhanouf, Rabah Wu, Yupeng
Keywords:	HVAC operations; Energy consumption; Thermal comfort; Equipment usage detection; Building energy performance; Deep learning
Subjects:	T Technology > TH Building construction > TH7005 Heating and ventilation. Air conditioning
Faculties/Schools:	UK Campuses > Faculty of Engineering > Built Environment
Item ID:	76665
Depositing User:	Wei, Shuangyu
Date Deposited:	29 Oct 2024 13:25
Last Modified:	29 Oct 2024 13:25
URI:	https://eprints.nottingham.ac.uk/id/eprint/76665

Actions (Archive Staff Only)

Edit View