Lithium-ion batteries have become a key component in the rise of electric mobility, but predicting their health and lifespan limits the technology.
Although tried and tested, the capacity of lithium-ion batteries degrades over time, and not only because of the aging process that occurs during charging and discharging – the so-called “cyclic aging”.
Batteries also deteriorate when not in use
Lithium-ion battery cells also suffer from what is known as “calendar aging” which occurs during storage or simply when the battery is not in use. It is determined by three main factors: the resting state of charge (SOC), the resting temperature, and the length of a battery’s resting time.
Hey EV nerd!
Subscribe now for a weekly roundup of our favorite mobility stories
Given that an EV will spend most of its lifetime parked, predicting cell capacity degradation due to calendar aging is critical. It can extend battery life and pave the way for mechanisms that could even circumvent the phenomenon.
Because of this, researchers have used advanced machine learning algorithms to accurately predict calendar aging.
Bridging research with the EV market
In a recent study funded by the EU’s Horizon 2020 program, a team of scientists went a step further by comparing the accuracy of two algorithms for the broad spectrum of commercial lithium-ion battery chemistry.
Specifically, they plotted calendar aging dates of six types of battery cell chemistries: lithium cobalt oxide (LCO), lithium iron phosphate (LIP), lithium manganese oxide (LMO), lithium titanium oxide (LTO), nickel cobalt Aluminum Oxide (NCA). and Nickel Manganese Cobalt Oxide (NMC).
These battery cells were calendar aged in high, medium and low voltage temperature chambers at 50, 60 and 70 degrees Celsius.
To predict aging, the team examined the efficiency of two machine learning algorithms: Extreme Gradient Boosting (XGBoost) and an Artificial Neural Network (ANN).
How do the algorithms work?
Both algorithms were chosen for their ability to provide reliable results, but they differ significantly in how they work.
XGBoost is a decision tree-based, state-of-the-art supervised machine learning algorithm commonly used in regression or classification problems.
The ANN is an artificial adaptive system that uses its basic elements, called neurons and connections, to transform its global inputs into a predicted output.
To assess their performance, the researchers used the mean absolute percentage error (MAPE) metric, which measures the average size of errors between predicted and measured values. Simply put, the smaller the MAPE value, the better the prediction accuracy.
What did the results show?
The tests of the algorithms showed that XGBoost can be used to effectively predict the calendrical aging of most chemistries with a significantly minimal mean absolute error. Meanwhile, ANN provides satisfactory results only for LFP, LTO, and NCA cell chemistry.
You can check their accuracy in the graph below:
Comparison of the accuracy of ANN and XGB based on how low the MAPE value is. Photo credit: B. Celen et al.
The overall superior performance of XGB, and in particular in relation to the battery chemistry that dominates in the automotive industry (NCA, NMC, LFP), shows that it could be integrated into EV battery application software to successfully predict calendar aging effects and provide longer service life for EV batteries .
It now remains to be seen what steps will have to be taken to translate the research results into commercial applications.
The full study can be found here.