A hybrid deep learning and clonal selection algorithm-based model for commercial building energy consumption prediction.

In contemporary society, commercial buildings, as a crucial component of urban development, face increasingly prominent energy consumption issues, posing significant challenges to the environment and sustainable development. Traditional energy management methods rely on empirical models and rule-based approaches, which suffer from low prediction accuracy and limited applicability. To address these issues, this study proposes a commercial building energy consumption prediction and energy-saving strategy model based on hybrid deep learning and optimization algorithms. This model integrates convolutional neural networks (CNN), gated recurrent units (GRU), and the clonal selection algorithm (CSA), aiming to enhance the accuracy and efficiency of energy consumption predictions. Experimental results demonstrate that the CNN-GRU-CSA Network (CGC-Net) model achieves mean absolute errors (MAE) of 17.12, 16.73, 16.62, and 15.94 on the Building Data Genome Project (BDGP), Commercial Building Energy Consumption Survey (CBECS), Nonresidential Building Energy Performance Benchmark (NEPB), and Building Energy Efficiency Benchmark (BEBDEE) datasets, respectively, significantly outperforming traditional methods and other models. Additionally, the model exhibits faster inference and training times. These results validate the stability and superiority of the CGC-Net model, providing an innovative solution and essential technical support for commercial building energy management.