Machine Learning for Adaptive Facade Design: Enhancing Thermal Performance in Urban Architecture

This study explores integrating machine learning models, specifically LightGBM and the Energy Valley Optimizer (EVO), in optimizing building façade designs for enhanced energy efficiency and thermal performance. LightGBM, a gradient boosting model, is utilized to predict thermal behavior and energy consumption by analyzing key architectural parameters such as façade type, orientation, and window-to-wall ratio (WWR). Meanwhile, the Energy Valley Optimizer (EVO), a nature-inspired metaheuristic algorithm, optimizes these parameters to balance energy efficiency with aesthetic and functional design requirements. The study demonstrates significant reductions in energy consumption across various building types, with composite façades achieving up to a 40% decrease in energy use compared to glass and metal alternatives. Optimized WWRs further reduced energy demand by 35 kWh/m^² in high-rise offices and 25 kWh/m^² in mid-rise residential complexes. Cross-climate analysis highlights the importance of adaptive, region-specific strategies, resulting in 20% to 30% improved energy efficiency in diverse climates. This research contributes a data-driven, climate-responsive framework for early-stage architectural design, bridging aesthetics, functionality, and sustainability. The study supports AI-driven strategies in achieving energy-efficient, high-performance buildings by advancing theoretical understanding and offering practical insights.