Comparative Carbon Footprint and Energy Analysis of Buildings Envelope Materials Based on Deep Learning Approach Prediction: Toward Sustainable Cities and Less Carbon Emission
Abstract
Proper prediction of energy use and carbon emission in urban buildings is very important in sustainable development and mitigation of environmental effects. In spite of the current developments of time‐series forecasting, current models can only capture nonlinear dependencies and converge to stable results with large and heterogeneous data. The proposed model presents a novel hybrid deep learning approach, based on the Long Short‐Term Memory (LSTM) and gated recurrent unit (GRU) networks, optimized through the Harris Hawks Optimization (HHO) algorithm, to improve predictive accuracy and parameter stability. The proposed model was applied on simulation data of the city of Riyadh, Saudi Arabia and Tashkent, Uzbekistan. The findings indicated strong accuracy in prediction of carbon emissions and energy consumption in the two cities. The findings showed that the material production was a leading source of carbon emission exceeding 85% of the total carbon emissions, and the total emission of Riyadh was almost twice that of Tashkent because of a greater electricity requirement to cool the place. The proposed model effectively captured the temporal dynamics of both energy use and emissions, providing a robust predictive tool for environmental assessment. The present study offers a sound predictive instrument of environmental analysis and urban energy planning of different climatic areas.