Paper Conference

Proceedings of Building Simulation 2021: 17th Conference of IBPSA


The impact of using the application of a CNN-based approach for equipment usage detection on building energy

Shuangyu Wei, Paige Wenbin Tien, Yupeng Wu, John Kaiser Calautit
University of Nottingham, United Kingdom

Abstract: The present study introduces an equipment usage detection approach using computer vision and deep learning methods for efficient building energy controls. The experimental results presented a detection accuracy of equipment detection of 83.33%. To investigate the impact of the proposed approach on building energy performance, the case study building was modelled and simulated. The simulation results showed that up to 35.95% lower internal heat gains was predicted with the use of deep learning influenced equipment detection profiles in comparison with the use of static or fixed schedules. The study highlights the benefits of incorporating real-time deep learning detection method with demand-driven controls which can minimize unnecessary building energy consumption while maintaining comfortable indoor environment.
Keywords: Deep learning, equipment detection, building energy savings, built environment, HVAC
Pages: 2411 - 2418