Smart Battery Management: Extending EV Battery Lifespan through Balancing and Machine Learning

Lithium battery packs are essential in electric vehicles (EVs). Differences in cell behavior and changing conditions can cause state of charge (SOC) imbalances, which are detrimental to battery performance and can result in structural degradation. In the experimental setup, improved charging and discharging mechanisms ensure SOC and SOH estimation in order to increase performance of the battery pack. With these improvements, there is a significant decrease in SOC with improved performance during the charging and discharging cycles. A battery life and remaining useful life (RUL) estimation can be performed using three different machine learning (ML) models: decision tree, k-nearest neighbor (KNN), and support vector machine (SVM). Of the different models, SVM has the best accuracy with the least error and greatest estimation performance. The proposed solution creates a closed feedback loop by SOC and SOH estimation focusing on increasing the battery pack's health for a longer period, while maintaining a predictive balance in monitoring battery health. The integrated solution is aimed at increasing the lifespan, reliability, and RUL of sustainable EVs and meeting the demand for sustainable transportation. [1]