Batteries in electric cars often age faster than other parts, which wastes resources and slows the shift to cleaner transport. To fix this, car makers are using software, often with artificial intelligence (AI) to better manage and control batteries. Scientists at Uppsala University have made a new AI model that can make battery health predictions up to 70 percent more reliable.
This model helps learn more about how batteries live and age, which can improve control systems in electric vehicles. It stresses the need to understand what happens inside batteries instead of treating them as black boxes, meaning unknown systems that just supply power. By getting a clear view of the inner processes, batteries can be handled to last longer.
Mapping the battery life cycle
The study used years of battery testing done with Aalborg University in Denmark. They built a database from data on many short charging segments, which are brief periods of charging. This data was mixed with a detailed model of the chemical processes inside the battery, like reactions that create power but also cause aging over time.
Together, this gives a sharp picture of the chemical reactions that let the battery work and how it wears out during use. The model can also spot safety issues in batteries, often from design problems or side reactions - unwanted chemical changes. These can be foreseen by looking at data from charging and discharging, when the battery gives out power.
The use of only short charging segments is a plus because full battery data from vehicles is sensitive for companies and user privacy. This shows good results can come without needing complete datasets. The work could lead to safer electric cars and less waste, helping the transport sector change faster.
The scientists have described the methods and results of this study in a paper published in Energy & Environmental Science.