IntelliCasa's research project presented at IAQ2020 international conference!
Updated: Jun 24
IntelliCasa is proud to announce that our research work titled “Analysis and Optimisation of Building Efficiencies through Data Analytics and Machine Learning” with the UCL EEE Department and General Technology Ltd has been presented at the international conference IAQ 2020: Indoor Environmental Quality Performance Approaches, on the 5th of May 2022. The conference was organised by ASHRAE and AIVC and many research papers from prestigious academic institutions and research centres globally have been presented in the areas of Indoor Environmental Quality (IEQ) in buildings. Moreover, our research paper will be published except the conference proceedings, also in the ASHRAE Science and Technology for the Built Environment (STBE) journal.
IntelliCasa has been involved in the “Building Data Analytics” research project that ignited as a result of the MOU signed with UCL University in 2018 among other activities. Konstantinos Karagiannis COO at IntelliCasa and MD at General Technology Ltd, proposed this idea to the UCL EEE department back in 2018, about utilising data analytics and machine learning models to improve the comfort levels of occupants and reduce operating costs and thus contributing to a more sustainable built environment. The project commenced as a collaboration between UCL (UK), General Technology Ltd (Greece) and IntelliCasa (UK). Konstantinos Karagiannis acted as an Industry advisor and research collaborator for the UCL research team, led by Dr Ryan Grammenos, EngD, SFHEA, Associate Professor and research students Hyunjee Kim, Changyou Gong and specifically Manuel Escalante Ruiz of the UCL Electronic and Electrical Engineering department. General Technology Ltd, a company specialising in building energy management systems provided the dataset of recorded building variables from a selected office building for the research and all research input needed for the completion of the project.
Building data analytics and machine learning models can contribute significantly to the transparent operation and performance of Smart buildings by identifying patterns of poor performance while eliminating occupant’s intervention in the system. These self-adaptive systems can predict occupant’s setpoint preference and balance the trade-off between thermal comfort and energy efficiency.
To this end, this work had the following two-fold aim; first, to develop a machine learning model to predict setpoint temperatures in an HVAC system towards the vision of making this system autonomous; second, to use exploratory data analysis techniques to evaluate the current operation and energy performance of an HVAC system in an office block by analysing and comparing patterns and trends between building management system (BMS) parameters and thermal comfort standards.
The machine learning model that was developed by the research team was able to predict the setpoint temperatures of an HVAC system in an office building and used data analytics to assess the thermal comfort and energy performance to balance this trade-off. The model managed to predict the user’s setpoint with 94.14% accuracy based only on environmental parameters, whilst adding time-related features accuracy was improved to 97.68%. Implemented exploratory data analysis techniques indicated that more than 50% of the room temperature values of the dataset lied outside the comfort zone for all seasons, while the setpoint deviation was equal to zero in only 4.31% of the cases, with an average value of 3.38°C (6.08°F). Additionally, analysis of the worst-case scenario of the hottest Summer day revealed that the HVAC system was underperforming since the desired setpoint was never reached, with a minimum difference of 3°C (5.4°F) and the room temperature kept rising for three hours before the system was turned off. Clearly, the current system was performing poorly in terms of thermal comfort and energy efficiency.
Conclusively, the results suggest that the implementation of a machine learning model not only will be able to predict the user’s setpoint but also improve the occupant's thermal comfort as well as the system's energy efficiency by optimisation, with benefits for both occupants and building owners.
Special thanks to Dr Ryan Grammenos for his excellent collaboration and significant comtribution to all stages of this research.