A Multimodal Physics-Informed Neural Network Approach for Mean Radiant Temperature Modeling

2025-03-11

AI Summary

Accurately estimating mean radiant temperature ($T_{mrt}$) for outdoor thermal comfort assessment in desert cities remains challenging due to complex radiation interactions and sparse in-situ measurements. To address this, we propose a physics-informed, multimodal modeling framework that integrates radiative transfer physics—specifically shortwave and longwave radiation equations—into a Physics-Informed Neural Network (PINN). The model jointly leverages fisheye-image-based shadow detection, real-time meteorological inputs, and built-environment descriptors, all under physical conservation constraints to ensure interpretability and generalizability. Evaluated on real-world desert urban sites, the method achieves an RMSE of 3.50°C and an $R^2$ of 0.88, significantly outperforming purely data-driven baselines. This work establishes a new paradigm for high-accuracy, lightweight, and real-time deployable dynamic assessment of urban thermal environments—bridging physical fidelity with deep learning efficiency.

Technology Category

• Computer Vision: Low Level & Physics-based Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web search models and ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Outdoor thermal comfort is a critical determinant of urban livability, particularly in hot desert climates where extreme heat poses challenges to public health, energy consumption, and urban planning. Mean Radiant Temperature ($T_{mrt}$) is a key parameter for evaluating outdoor thermal comfort, especially in urban environments where radiation dynamics significantly impact human thermal exposure. Traditional methods of estimating $T_{mrt}$ rely on field measurements and computational simulations, both of which are resource intensive. This study introduces a Physics-Informed Neural Network (PINN) approach that integrates shortwave and longwave radiation modeling with deep learning techniques. By leveraging a multimodal dataset that includes meteorological data, built environment characteristics, and fisheye image-derived shading information, our model enhances predictive accuracy while maintaining physical consistency. Our experimental results demonstrate that the proposed PINN framework outperforms conventional deep learning models, with the best-performing configurations achieving an RMSE of 3.50 and an $R^2$ of 0.88. This approach highlights the potential of physics-informed machine learning in bridging the gap between computational modeling and real-world applications, offering a scalable and interpretable solution for urban thermal comfort assessments.

Problem

• Modeling Mean Radiant Temperature for urban thermal comfort
• Overcoming resource-intensive traditional estimation methods
• Integrating physics and deep learning for accurate predictions

Innovation

• Physics-Informed Neural Network for $T_{mrt}$ modeling
• Multimodal dataset integration for predictive accuracy
• Combines deep learning with radiation physics