Prediction and performance evaluation of building-integrated photovoltaic facades using artificial intelligence models

Abstract

The continuous growth of global energy consumption due to industrialization and urban development has significantly contributed to climate change. In response, renewable energy, particularly solar energy, has emerged as a key solution toward green and sustainable building design. This study proposes the application of artificial intelligence algorithms, including Gradient boosting (GB), Extreme gradient boosting (XGBoost), Light gradient boosting machine (LightGBM), and Categorical boosting (CatBoost), to develop predictive models for building energy consumption and photovoltaic electricity generation in office buildings equipped with glass-based building-integrated photovoltaics (BIPV) facades. Input data were generated through simulation using DesignBuilder software. Among the models, CatBoost yielded the best performance, with R² values of 0.972 for energy consumption and 0.974 for electricity generation. Furthermore, the integration of BIPV systems into building facades has the potential to reduce energy consumption by up to 34.3%, highlighting its practical benefits for energy-efficient building design.