Sensorless Temperature Estimation for Li-Ion Battery Cells: An Overview, Practical Considerations, Challenges and Future Trends

Sensorless temperature estimation methods for batteries can be classified into three categories: analytical methods, observer-based methods and data-driven methods. In general, analytical methods are easy to derive and implement but have a limited performance due to their open-loop nature. Observer-based methods have a high performance due to their closed-loop nature but demand accurate dynamic and measurement models with up-to-date parameters for accurate estimation. Data-driven methods are extremely stable and robust due to their massive parallel structure, but they demand huge amount of data for training. This paper presents a comprehensive review of state-of-the-art sensorless temperature estimation methods for lithium-ion (Li-ion) batteries and demonstrates the advantages and limitations of each. A comparison between the presented methods in terms of their performance and implementation requirements is carried out. In addition, practical considerations, challenges and future trends are discussed.