Enhancing electric vehicle battery lifespan: integrating active balancing and machine learning for precise RUL...

Area of Science:

• Electrical Engineering
• Materials Science
• Computer Science

Background:

• Electric vehicles (EVs) utilize lithium-ion battery packs for energy storage.
• Inconsistent cell characteristics and operating conditions cause imbalanced state of charge (SOC) levels, reducing EV battery capacity and accelerating degradation.
• Effective battery management systems are crucial for optimizing EV performance and longevity.

Purpose of the Study:

• To develop and validate an active cell balancing method for lithium-ion battery packs in EVs.
• To enhance both charging and discharging capacities by equalizing SOC across cells.
• To evaluate machine learning models for accurate prediction of battery lifespan and remaining useful life (RUL).

Main Methods:

• Implementation of a two-strategy active cell balancing system: charging balance for capacity maximization and discharging balance for duration extension.
• Evaluation of seven machine learning models (k-nearest Neighbors, Random Forest, etc.) using R-squared (R²) and Mean Absolute Error (MAE) for RUL prediction.
• Integration of active balancing with RUL prediction to create a feedback loop for optimized battery management.

Main Results:

• The proposed active balancing method effectively reduced SOC disparities, leading to enhanced charging and discharging capacities.
• K-nearest Neighbors and Random Forest models demonstrated high accuracy in RUL prediction, achieving R² values of 0.996+ and low MAE.
• The integrated system showed that balanced SOC levels promote battery health, while RUL predictions inform optimal balancing strategies.

Conclusions:

• Active cell balancing is vital for improving the performance and lifespan of EV lithium-ion battery packs.
• Machine learning offers a powerful tool for accurate battery RUL prediction, crucial for proactive maintenance.
• The synergistic integration of active balancing and RUL prediction represents a significant advancement in EV battery management systems, enhancing reliability and longevity.