It is quite difficult to see into the mounted traction battery, but AI can help detect potential problems.
Due to the “newness” of electric vehicles and their difficulty in extinguishing, electric car fires are a popular topic for various sensationalist media and individuals. However, they happen very rarely. A new technique developed by Eatron Technologies, part of the large Hirschvogel automobile concern, could now reduce the risk of fires. “As an industry, we need to ensure that the number of catastrophic battery failures reaches zero and then stays there. Our intelligent, connected and secure automotive battery management software has shown that artificial intelligence is the key to achieving this goal.” stated Dr. Umut Genc, CEO of Eatron Technologies.
The causes of battery failure are complex and often involve a combination of factors. One of the most common causes – lithium plating – occurs when metallic lithium deposits form around the anode. This happens most often during fast charging at low temperatures, then during extreme discharge or continuous charging of the battery to maximum voltage. Over time, these deposits reduce the battery’s performance and can lead to the growth of dendrites, needle-like structures that can break through the separator between the anode and cathode and cause a short circuit in the cell. This in turn leads to rapid self-discharge, which can initiate a thermal discharge, a self-sustaining chain reaction that is difficult to quench.
Determining the lithium plating without opening the battery cell and examining the electrodes – which is essentially impossible after installation in the vehicle – is a challenge that is the subject of intensive research. And although various techniques have been developed over the years, each has its limitations, especially when it comes to distinguishing lithium plating from other degradation mechanisms. However, Eatron Technologies has proven with artificial intelligence that it is possible not only to detect lithium plating much more effectively, but also to accurately predict when it may occur.
“Using a technique called feature extraction, we transform the raw data coming from the battery into a format that makes it easier to identify anomalies. Combined with our proprietary AI pipeline that accurately captures battery behavior, our AI diagnostics can predict cell failure before it happens with up to 90% accuracy and zero false positives.” explained Dr. Umut Genc. Thanks to an early warning, it is then possible to direct the vehicle owner to a service center and solve the potential problem before a fire occurs.
