AUTHOR=Ma Chunying , Xu Yixiong 

TITLE=Research on construction and management strategy of carbon neutral stadiums based on CNN-QRLSTM model combined with dynamic attention mechanism

JOURNAL=Frontiers in Ecology and Evolution

VOLUME=Volume 11 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/ecology-and-evolution/articles/10.3389/fevo.2023.1275600

DOI=10.3389/fevo.2023.1275600

ISSN=2296-701X

ABSTRACT=As a large-scale construction project, sports stadiums are typically associated with significant energy consumption and carbon emissions. Therefore, promoting the development of carbonneutral stadiums and adopting sustainable strategies has become a pressing issue. This study aims to explore an integrated approach to construction and management, combining advanced convolutional neural networks (CNN) and quasi-recurrent long short-term memory (QRLSTM) with dynamic attention mechanisms, to achieve the sustainable development of carbon-neutral stadiums. The CNN-QRLSTM model, which combines CNN and QRLSTM characteristics, along with dynamic attention mechanisms, will be employed. The strengths of the CNN-QRLSTM model lie in its ability to handle both image and sequential data, making it suitable for the diverse information involved in stadium construction and management. The dynamic attention mechanism will enable the model to adaptively adjust attention weights based on different situations, capturing relevant information more accurately. By comprehensively applying deep learning models and dynamic attention mechanisms, we aim to provide more scientific decision support to stadium constructors and managers, helping them make sustainable choices in carbon reduction and resource utilization. We conducted experiments on four data sets: EnergyPlus, ASHRAE, CBECS and UCI. The experiments showed that compared with other advanced models, our proposed model achieved the highest score (97.79%) in terms of accuracy, recall rate, F1 In terms of scores and AUC, our method also achieves the highest scores.