With the increase of production of electrical vehicles (EVs) and battery packs, lithium ion batteries inconsistency problem has drawn much attention. Lithium ion battery imbalance phenomenon exists during three different stages of life cycle. First stage is premanufacturing of battery pack i.e., during the design, the cells of similar performance need to be clustered to improve the performance of pack. Second is during the use of battery pack in EVs, batteries equalization is necessary. In the third stage, clustering of spent lithium ion batteries for reuse is also an important problem because of the great recycling challenge of lithium batteries. In this work, several clustering and equalization methods are compared and summarized for different stages. The methods are divided into the traditional methods and intelligent methods. The work also proposes experimental combined clustering analysis for new lithium-ion battery packs formation with improved electrochemical performance for electric vehicles. Experiments were conducted by dismantling of pack and measurement of capacity, voltage, and internal resistance data. Clustering analysis based on self-organizing map (SOM) neural networks is then applied on the measured data to form clusters of battery packs. The validation results conclude that the battery packs formed from the clustering analysis have higher electrochemical performance than randomly selected ones. In addition, a comprehensive discussion was carried out.
Real-Time State-of-Health Estimation of Lithium-Ion Batteries Based on the Equivalent Internal Resistance
