An AI fast charging battery strategy has been developed by researchers at Chalmers University of Technology in Sweden that extends electric vehicle battery life by approximately 23 percent without increasing charging time. The new method adapts the charging current to the battery’s chemistry and state of health, reducing long-term degradation while maintaining the speed of conventional fast charging systems.
Changfu Zou, professor at the Department of Electrical Engineering at Chalmers, and Meng Yuan, Assistant Professor at Victoria University of Wellington, New Zealand, published their findings in IEEE Transactions on Transportation Electrification. The research shows that current can be adjusted during each fast charge to account for the battery’s age and chemical composition, mitigating harmful reactions that occur under standard charging protocols.
The Battery Degradation Challenge
Electric vehicle batteries currently have a lifespan of approximately 8 to 15 years, depending on use and charging patterns. Fast charging is necessary for longer journeys and is particularly important for taxi services, commercial vehicles, and passenger cars where availability outside the home enables commuting over extended distances. However, this convenience comes with a tradeoff: fast charging stresses batteries and reduces their operational lifespan.
When batteries are charged rapidly, a large current forces into various cells, increasing the risk of chemical side reactions. One significant problem is lithium plating, in which metallic lithium deposits on the electrode instead of storing correctly within the battery structure. This degradation reduces capacity and may affect safety, as uneven lithium distribution can potentially cause short circuits. The risk of lithium plating increases as batteries age, yet standard charging methods use identical current and voltage regardless of whether a battery is new or has been in use for years.
How AI Adapts the Charging Process
The new artificial intelligence strategy relies on reinforcement learning, a machine learning method in which algorithms improve decisions based on feedback. The AI model was trained using a simulation of one of the most common electric vehicle batteries on the market, learning to adapt charging according to the battery’s charge level at the time of charging and its overall state of health.
The result is a charging strategy that maintains short charging times while minimizing harmful electrochemical reactions. Meng Yuan stated that smart adaptation of current during charging, taking into account the changing electrochemical state of the battery, maximizes both performance and lifespan. The method requires no new hardware; implementation is possible through software updates to vehicle battery management systems.
Implementation and Industry Impact
The charging strategy is cost-effective to implement, though some calibration is needed for broader use across different battery types. The researchers plan to use transfer learning to adapt the AI model to new battery chemistries more rapidly. For the automotive industry, a 23 percent increase in battery life translates to lower warranty costs, improved resale value, and more efficient use of critical raw materials used in battery production.
The next phase involves testing the method directly on physical batteries. Researchers anticipate that this AI-based charging strategy will contribute significantly to electrification of the transport sector. Meng Yuan emphasized that fast charging combined with increased battery life are important factors encouraging people to switch to electric vehicles as part of the transition to a fossil-free society.
Future Outlook
The study was supported by the European Union’s Horizon Europe research program, the Swedish Research Council, and the Swedish Foundation for International Cooperation in Research and Higher Education. As the research moves from simulation to physical testing, the potential for widespread adoption depends on successful calibration across the diverse battery types currently in use by vehicle manufacturers globally.
