A review on machine learning driven next generation thermoelectric generators

Thermoelectric (TE) materials and devices have emerged as promising technologies for energy conversion and waste heat recovery. Yet, challenges remain in improving TE efficiency due to the complicated relationship between TE material properties, geometrical designs, and performance optimization. Machine learning (ML) on the other hand has played a great role in discovery and optimization of TEMs, offering new insights and accelerating the development of high-performance TE generators (TEG). This review provides a comprehensive overview of the applications of ML in TEG research, highlighting the methodologies and techniques employed for performance prediction and optimization of various TEGs including TEGs, segmented (STEGs), annular (ATEGs), photovoltaic (PVTEGs), and other hybrid (HTEGs) along with their maximum power point tracking (MPPT) and integration of TEGs with other systems. Key advancements include the use of regression and optimization as well as physics informed methods. Also, the potential benefits of integrating ML methods with other computational schemes such as COMSOL and CFD are highlighted. This review also addresses the challenges of limited datasets, model interpretability, and experimental validation. Future research directions are proposed, focusing on integrating ML with experimental and computational approaches to unlock new pathways for thermoelectric generators’ design. In general, this review provides valuable insights and potential of AI/ML into the advancement of TEG technology.