A reinforcement-learning–based charging approach from Sweden’s Chalmers University of Technology suggests artificial intelligence can enable faster EV charging while significantly extending battery life.
Electric vehicles are becoming more common as they reduce reliance on traditional internal-combustion engines, but battery life remains a key concern as demand for fast charging grows. Rapid charging imposes greater stress on lithium‑ion batteries, which can accelerate aging and degrade performance over time. A 2024 Geotab study indicated average annual battery degradation around 1.8%. For Tesla, estimates suggest batteries may operate between 300,000 and 500,000 miles depending on operating conditions and fast-charging habits.
The study describes a battery-management system that uses reinforcement learning to adjust charging current throughout the fast-charging process. Rather than applying a fixed protocol for all packs, the AI continuously analyzes data and adapts the charging strategy to each battery’s real‑time state. After each charging session, the algorithm learns from new data to refine the current-to-charging plan, aiming to reduce degradation while preserving charging speed. Meng Yuan, a postdoctoral researcher in the Electrical Engineering Department at Chalmers, said this marks the first time the team has formulated a clear expression for the fast-charging problem across the full battery life cycle. The method reportedly achieves a substantial improvement by extending battery life by about 703 full-charge cycles, roughly a 22.9% increase over standard approaches.
If validated in real-world operation, the approach could enable longer-lasting batteries without sacrificing the fast charging experience, potentially lowering ownership costs and increasing consumer confidence in electric mobility. Beyond hardware changes, future manufacturers could leverage AI to optimize how the battery operates throughout its life, without altering material structures. Real-world validation remains necessary, as current results are based on models and simulations rather than everyday use of commercial packs.
Meng Yuan, a postdoctoral researcher at Chalmers, said the work represents the first clear formulation of the fast-charging problem across the full battery life cycle. The study, published in IEEE, notes that while the results are promising, real-world validation is still required before deployment in commercial packs.
