Integrating Generative AI into Predictive Modeling for Energy Efficiency Optimization in Building Design

Buildings account for a notable portion of global energy consumption, making energy efficiency critical for sustainable development. Reducing heating and cooling loads is essential for improving the energy efficiency of buildings. Advances in machine learning have introduced robust methods for predicting energy consumption, enabling designers and engineers to optimize building designs. Recently, generative artificial intelligence (AI) has emerged as a robust tool for generating new, optimized designs that maximize energy efficiency by simulating building designs. Although predictive modeling has been widely applied to estimate heating and cooling loads, the challenge remains to design buildings that inherently reduce these energy demands. Traditional methods of architectural design often overlook optimization from an energy efficiency perspective, and manual adjustments based on predictions are inefficient. There is a need for intelligent systems that not only predict energy consumption but also generate optimized building designs that address energy inefficiency at the root level. In this study, we trained a random forest (RF) regressor to predict heating and cooling loads based on building features, such as relative compactness, surface area, wall area, and roof area. The RF model was trained and tested using a dataset of building features, achieving high predictive accuracy. In addition, a generative adversarial network (GAN) was applied to generate new building designs that optimize energy efficiency. The RF model was further used to evaluate the energy efficiency for the generated designs, offering a comprehensive approach to both prediction and optimization. The RF model demonstrated excellent performance in predicting heating and cooling loads, with training and testing mean squared errors of 0.2947 and 1.7946, with corresponding R² values of 0.9967 and 0.9802, respectively. Feature importance analysis revealed that relative compactness and overall height were the most significant factors influencing energy consumption. Through the integration of generative AI, we generated new building designs that were tested for energy efficiency, showing a marked improvement in heating and cooling load reductions. The results demonstrate that combining predictive models with generative AI offers a powerful solution for optimizing energy efficiency in building designs. Using GANs, architects and engineers can automatically generate designs that minimize energy consumption while adhering to structural and aesthetic requirements. This approach represents a promising advancement in sustainable architecture and energy-efficient building design, paving the way for smart, data-driven design processes that align with global sustainability goals.