Machine learning assisted advanced battery thermal management system: A state-of-the-art review

Abstract

With an increasingly wider application of the lithium-ion battery (LIB), specifically the drastic increase of electric vehicles in cosmopolitan cities, improving the thermal and fire resilience of LIB systems is inevitable. Thus, in-depth analysis and performance-based study on battery thermal management system (BTMs) design have arisen as a popular research topic in energy storage systems. Among the LIB system parameters, such as battery temperature distribution, battery heat generation rate, cooling medium properties, electrical properties, physical dimension design, etc., multi-factor design optimisation is one of the most difficult experimental tasks. Computational simulations deliver a holistic solution to the BTMs design, yet it demands an immense amount of computational power and time, which is often not practical for the design optimisation process. Therefore, machine learning (ML) models play a non-substitute role in the safety management of battery systems. ML models aid in temperature prediction and safety diagnosis, thereby assisting in the early warning of battery fire and its mitigation. In this review article, we summarise extensive lists of literature on BTMs employing ML models and identify the current state-of-the-art research, which is expected to serve as a much-needed guideline and reference for future design optimisation. Following that, the application of various ML models in battery fire diagnosis and early warning is illustrated. Finally, the authors propose improved approaches to advanced battery safety management with ML. This review paper aims to bring new insights into the application of ML in the LIB thermal safety issue and BTMs design and anticipate boosting further advanced battery system design not limited to the thermal management system, as well as proposing potential digital twin modelling for BTMs.