Machine Learning Enhancement of Heat Transfer Research and Technology Development

This review has two main threads of discussion. The first describes key underlying foundational concepts of the machine learning (ML) tools considered. This is included for two reasons. First, these concepts often dictate how the data science strategy interacts with the physics of the system, which is key to creating robust, high-fidelity physics-inspired models, and key to adapting ML tools to different systems and objectives. In addition, the discussion of the fundamental data science principles behind ML tools is included to make this review useful to a graduate student in engineering with little prior experience who is interested in using ML in their heat transfer research. The second main discussion thread summarizes recent research (as of 2025) exploring the use of ML tools in heat transfer and energy conversion research and development. This discussion focuses on studies involving physics-inspired ML tools that have been most widely used in heat transfer−related applications: artificial neural networks, genetic algorithms, convolution neural networks, and physics-inspired neural networks. Research on