Abstract
<jats:p>Accurate prediction of heat transfer is essential for the design and optimization of thermal systemsoperating under complex and nonlinear conditions. Traditional differential equation–based heattransfer models provide strong physical interpretability but often require high computational effortand simplifying assumptions. This study presents a machine learning–supported framework for heattransfer prediction that integrates physics-based differential equation models with data-driven learningtechniques. Numerical solutions of governing heat transfer equations are used to generate reliablethermal datasets, which are then employed to train machine learning models capable of capturingnonlinear thermal behavior. A hybrid approach combining machine learning predictions with physics-based constraints is developed and evaluated. Comparative results demonstrate that the hybrid modelachieves high prediction accuracy while significantly reducing computational time when compared toconventional numerical methods. The proposed framework offers a reliable and efficient solution foradvanced heat transfer prediction in engineering applications.</jats:p>