A research team has developed a hybrid method combining physical models with machine learning to infer key degradation parameters in lithium-ion batteries. This approach allows for more precise and efficient identification of the underlying causes of battery capacity deterioration than traditional methods. The ability to understand and predict degradation is crucial for optimizing the design, management, and lifespan of these batteries, which are fundamental in electric vehicles and electronic devices.

The study addresses the complexity of lithium-ion battery degradation, which involves multiple interconnected mechanisms such as solid-electrolyte interphase (SEI) growth, active material loss, and particle fracture. Traditionally, identifying these parameters requires destructive experiments or complex analyses unsuitable for real-time monitoring. The new methodology uses a physics-based battery model to simulate behavior, then employs machine learning techniques to fit model parameters to experimental degradation data, achieving inverse inference of the dominant mechanisms.

This advance could have a significant impact on the battery industry. By being able to diagnose more accurately why and at what rate a battery degrades, manufacturers could design more durable and efficient batteries. Furthermore, this technique could be used to develop smarter battery management systems that optimize charge and discharge cycles to extend lifespan, or to predict failures before they occur, improving the safety and reliability of devices and vehicles that use them.