Machine Learning Models for Predicting Thermal Properties of Radiative Cooling Aerogels

Area of Science:

• Materials Science
• Climate Science
• Artificial Intelligence

Background:

• The urgent need for efficient cooling materials is driven by the global climate crisis and energy demands.
• Radiative cooling materials offer a passive solution for reducing energy consumption and mitigating heat island effects.
• Developing high-performance radiative cooling aerogels (RCAs) requires understanding complex material-environment interactions.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for predicting the cooling performance of radiative cooling aerogels (RCAs).
• To identify key material and environmental parameters influencing RCA performance.
• To provide theoretical guidance for optimizing RCA design through feature analysis.

Main Methods:

• Integration of diverse parameters including material composition, modification design, optical properties, and environmental factors into an ML model.
• Comparative analysis of various ML algorithms, with a focus on XGBoost for its predictive accuracy.
• Application of Shapley Additive Explanations (SHAPs) for model interpretability and feature importance analysis.

Main Results:

• An optimized XGBoost model achieved high predictive accuracy for RCA performance, with R² = 0.943 and RMSE = 1.423 on the test dataset.
• SHAP analysis identified zinc oxide (ZnO) as a significant modifier, and ambient temperature and solar irradiance as critical environmental determinants of cooling performance.
• Feature interaction analysis revealed complex relationships between material composition and environmental conditions affecting cooling efficiency.

Conclusions:

• Machine learning, particularly the XGBoost model, provides a robust framework for predicting RCA performance.
• ZnO modification and specific environmental conditions are crucial factors for optimizing radiative cooling efficiency.
• The developed model and analyses offer valuable insights for the rational design of advanced radiative cooling materials to address climate and energy challenges.