Computer scientists at Loughborough University have collaborated with engineering consultancy Cundall on developing an artificial intelligence (AI) system that can rapidly predict building energy rates.
Building emissions rates (BER) are an important component of calculating energy performance and efficiency in buildings, and are required for the completion of a building’s energy performance certificate (EPC).
Current methods of producing BERs are generated by manually inputting hundreds of variables, which can take hours to days to generate depending on a building’s complexity.
A research team led by Dr Georgina Cosma and postgraduate student Kareem Ahmed, of Loughborough’s School of Science, now claims to have designed and trained an AI model that can predict BER values of non-domestic buildings in less than a second, with as few as 27 variables with little loss in accuracy.
Dr Cosma described the research as ‘an important first step towards the use of machine learning tools for energy prediction in the UK’, showing how data can ‘improve current processes in the construction industry’.
Created with support from Cundall’s head of research and innovation Edwin Wealend, the AI model was reportedly trained using large-scale data obtained from UK government energy performance assessments.
Researchers said they used a ‘decision tree-based ensemble’ machine algorithm and built and validated the system using 81,137 real data records containing information for non-domestic buildings, such as shops, offices and restaurants, across England from 2019 to 2019. The data contained information including building capacity, location, heating, cooling, lighting and activity.
“Studies on the applications of machine learning on energy prediction of buildings exist, but these are limited, and even though they only make up eight per cent of all buildings, non-domestic buildings account for 20 per cent of UK’s total CO2 emissions,” said Dr Cosma.
Wealend added that the team eventually hopes to build on the techniques developed to predict real operational energy consumption.
“By predicting the energy consumption and emissions of non-domestic buildings quickly and accurately, we can focus our energy on the more important task — reducing energy consumption and reaching Net Zero,” he commented.
Project findings were presented at The Chartered Institution of Building Services Engineers (CIBSE) Technical Symposium 2021 and the paper will be published on CIBSE’s website later this year.