Deep Learning Driven Engineering Innovations: Advancing Sustainable Green Building and Smart Infrastructure for a Resilient Future

With increasing global focus on sustainability and energy efficiency, green building technologies and smart infrastructure have emerged as key solutions to address environmental concerns and resource scarcity. In this study, we present a hybrid deep learning model combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks to enhance predictive accuracy for smart infrastructure and green building management. The proposed model utilizes CNN layers to extract spatial patterns from sensor data (e.g., temperature, energy consumption across building zones) and LSTM layers to process time-series data (e.g., occupancy and energy usage trends). The model addresses both spatial and temporal dimensions of building performance data, ensuring accurate predictions of energy consumption and optimizing environmental conditions. The model is trained using a comprehensive dataset from smart building systems, and its performance is evaluated against nine other existing models, including Logistic Regression, Decision Trees, Support Vector Machines, Random Forest, Gradient Boosting, XGBoost, LightGBM, CNN, and LSTM. The proposed hybrid model outperforms these models, achieving an accuracy of 97.65%, with the lowest Mean Squared Error (MSE) of 0.0215 and Mean Absolute Error (MAE) of 0.0184. This improved performance demonstrates the hybrid model's capability to predict building energy needs efficiently and enhance decision-making in smart infrastructure. These results are vital for advancing sustainable green buildings and optimizing energy usage in real-time.