A Morphological Generative Framework for Climate-Adaptive Building-Integrated Photovoltaics (BIPV) Facades Integrating Artificial Intelligence Algorithms and Bayesian Prior-Parameterized Building Envelopes

The flexible and precise design of photovoltaic (PV) skin morphologies for building facades constitutes a critical element for optimizing building energy efficiency and enhancing indoor spatial performance. However, the complex interrelationships among climatic parameters and their spatiotemporal variations pose significant challenges to the prior validity and accuracy of climate-adaptive parametric skin morphology adjustments. To address these limitations, this study proposes a morphological generative framework for climate-adaptive Building-Integrated Photovoltaics (BIPV) facades integrating Artificial Intelligence Algorithms and Bayesian prior-parameterized building envelopes. This framework is specifically designed to facilitate morphological decision-making regarding the overall climate-adaptive opening states of parametric PV skins under spatiotemporal dynamics. The proposed method integrates AI-based pattern recognition in spatiotemporal climate data with Bayesian Network-based prior probability techniques to derive optimal facade morphology schemes with the highest overall climate adaptability scores derived from weather forecasts, thereby achieving optimal transformations of the building envelope. Specifically, the model first employs an Artificial Intelligence Algorithm to generate the Bayesian Network structure required for overall climate adaptability scoring. Secondly, utilizing the Chinese Standard Weather Data (CSWD), the GRASSHOPPER algorithm is applied to implement variable parametric design on the facade skin, generating dynamic parametric skins and visual climatic data analysis cloud maps for energy benefit assessment. Finally, facade updates are executed based on the overall climate adaptability scores. The results demonstrate that the proposed framework effectively enables the real-time selection of optimal morphologies and opening states for dynamic skins based on comprehensive climatic adaptability criteria. Following model training and validation using 2025 Panjin meteorological data in the EnergyPlus Weather (EPW) format, the generated facade morphologies yielded solar radiation gains of 166.9 kWh/m2·month (peak month) for one of the optimal summer configurations and 90.5 kWh/m2·month (December) for one of the optimal winter configurations. Furthermore, by providing definitive evaluations of PV energy yields and indoor comfort levels across diverse weather scenarios, this framework offers explicit guidance for skin design, thereby reconciling the multi-objective optimization relationship between building energy conservation and occupant comfort.