AI Can Sort Batteries

A new machine-learning process can be used to predict lithium ion battery cycle life, based upon factors measured in just a few early charge and discharged cycles.

Although commercial lithium ion battery production is a highly automated process, there is still significant variability in the product that comes off the assembly line. Typically, the capacity of a lithium-ion battery is stable for a while before sharply decreasing. The point at which the decrease occurs varies widely. Predicting which batteries will last the longest is problematic but If manufacturers could tell which cells will last the longest they could send those cells to more critical applications (cell phones, for example) and send the rest to be used in less demanding devices.

Now scientists at Stanford University, the Massachusetts Institute of Technology and the Toyota Research Institute have developed a technique using artificial intelligence (AI) and a machine learning process to predict how many cycles each battery would last, based on voltage declines and a few other factors among the early cycles. The machine learning method could also be used to accelerate research and development of new battery designs, according to information provided by a Stanford University news release.

Stanford graduate students Nicholas Perkins, left, Peter Attia and Norman Jin are among the researchers who found the key for accurately predicting the useful life of lithium-ion batteries. (Image source: Dean Deng)

Removing a Bottleneck

“The standard way to test new battery designs is to charge and discharge the cells until they fail. Since batteries have a long lifetime, this process can take many months and even years,” said co-lead author of the study Peter Attia, Stanford doctoral candidate in materials science and engineering. “It’s an expensive bottleneck in battery research.”

According to the news release, “After the researchers trained their machine learning model with a few hundred million data points of batteries charging and discharging, the algorithm predicted how many more cycles each battery would last, based on voltage declines and a few other factors among the early cycles.”

The predictions made by the algorithm were within 9 percent of the number of cycles the cells actually attained. The algorithm was able to categorized batteries as either long or short life expectancy based on just the first five charge/discharge cycles, and the predictions were correct 95 percent of the time. The researchers have made the dataset—the largest of its kind—publicly available.

Many Potential Applications

The news release from Stanford states, “The new method has many potential applications. For example, it can shorten the time for validating new types of batteries, which is especially important given rapid advances in materials. With the sorting technique, electric vehicle batteries determined to have short lifespans—too short for cars—could be used instead to power street lights or back up data centers. Recyclers could find cells from used EV battery packs with enough capacity left for a second life.”

MIT’s team in the project, was led by Richard Braatz, professor in chemical engineering and his group performed the machine learning work. “Advances in computational power and data generation have recently enabled machine learning to accelerate progress for a variety of tasks. These include prediction of material properties,” Braatz said. “Our results here show how we can predict the behavior of complex systems far into the future.”

Optimizing Manufacturing

Yet another possibility is optimizing battery manufacturing. “The last step in manufacturing batteries is called ‘formation,’ which can take days to weeks,” Attia said. “Using our approach could shorten that significantly and lower the production cost.”

During formation the battery cell must be put through at least one precisely controlled charge and discharge cycle which is used to activate the working material. Instead of using the typical normal constant current-constant voltage charging curve, during formation the charging begins with a low voltage, which builds up gradually. The process is used to create the SEI (solid electrolyte interface) on the battery anode.

Data such as capacity and impedance are gathered and recorded for quality analysis during the formation stage, and gives some indication of whether or not the manufacturing process is under control. It should also be able to cull out cells that would be expected to fail early in their life.

Using the new dataset and AI to move battery testing forward is a big step. “For all of the time and money that gets spent on battery development, progress is still measured in decades,” said Patrick Herring, a scientist at the Toyota Research Institute and who was a part of the research team. “In this work, we are reducing one of the most time-consuming steps—battery testing— by an order of magnitude.”

Senior Editor Kevin Clemens has been writing about energy, automotive, and transportation topics for more than 30 years. He has masters degrees in Materials Engineering and Environmental Education and a doctorate degree in Mechanical Engineering, specializing in aerodynamics. He has set several world land speed records on electric motorcycles that he built in his workshop.

 

logo   

                                                  September 10-12, 2019

                                                   CALL FOR Speakers! 

   SUBMISSION DEADLINE: Friday, March 29, 2019, 11:59 PM EST

 

Comments (0)

Please log in or to post comments.
  • Oldest First
  • Newest First
Loading Comments...