The merging of machine learning and battery research is reshaping how we perceive battery performance and longevity while also leaving a profound mark on patent trends.
In 2022, about 60% of lithium, 30% of cobalt, and 10% of nickel demand was for EV batteries. Just five years earlier, in 2017, these shares were around 15%, 10%, and 2%, respectively.
The amalgamation of machine learning with battery research offers exciting prospects for innovation and safeguarding intellectual property. However, it also brings forth fresh challenges in patent filing and legal disputes. This article explores these dynamics, offering researchers an idea of the current landscape of predictive battery performance.
The Impact of Machine Learning on Predictive Battery Performance
The scope of new research and developments has been constrained earlier. But today, with the evolution of sophisticated algorithms and a significant increase in computing power, a drastic change has taken place.
Let’s explore some of the most noteworthy examples illustrating this shift:
1. National Renewable Energy Laboratory (NREL)
The National Renewable Energy Laboratory (NREL) has been at the forefront of integrating machine learning into battery research. They use machine learning to characterize battery performance, lifetime, and safety. This is an innovative approach that helps in understanding new materials, chemistries, and cell designs.
Machine Learning Algorithm for Battery
One of the significant achievements of NREL is the development of a predictive battery performance algorithm that “rediscovered” relevant physical equations similar to those used by experts but without prior assumptions.
This was accomplished by applying machine learning algorithms to automatically generate thousands of equation components and down-select from millions of possible combinations to identify a parsimonious model that balances predictive accuracy with simplicity.
This machine learning model was able to predict a 40% to 130% longer calendar life when extrapolated forward in time, dependent on the aging condition.
This is a significant improvement over traditional methods and showcases the potential of machine learning in advancing battery research.
Leveraging machine learning with EBSD
In addition to this, NREL also used machine learning in combination with electron backscatter diffraction (EBSD) to map the orientation and morphology of sub-particle grains in 3D. When combined with machine learning image segmentation techniques, EBSD provides realistic 3D particle architectures for multi-physics modeling.
This combined approach represents the first demonstration of mapping and simulating dynamic phenomena within single electrode particles.
2. Stanford University, the Massachusetts Institute of Technology, and the Toyota Research Institute
A team of scientists from Stanford University, the Massachusetts Institute of Technology, and the Toyota Research Institute on March 25, 2019, discovered a method to accurately predict the useful life of lithium-ion batteries before their capacities start to wane.
This work was conducted at the Centre for Data-Driven Design of Batteries, an academic-industrial partnership that combines theory, experiments, and data science. The Stanford researchers carried out the battery tests under the direction of William Chueh, an assistant professor of materials science and engineering. The machine learning study was carried out by a team from MIT under the direction of Professor Richard Braatz of chemical engineering.
Revolutionizing battery lifespan prediction with AI
Combining comprehensive experimental data and artificial intelligence, the researchers trained their machine-learning model for predictive battery performance with a few hundred million data points of battery charging and discharging. The algorithm then predicted how many cycles each battery would last based on voltage declines and other factors among the early cycles.
The algorithm categorized batteries as long or short life expectancies based on the first five charge/discharge cycles. Here, the predictions were correct 95% of the time.
Among its many uses, this machine-learning technique might expedite the creation of new battery designs and cut production time and costs. The researchers have made the dataset – the largest of its kind – publicly available.
Recent Patents in Predictive Battery Performance
Here are some of the patents regarding predictive battery performance:
- Artificial Intelligence-Based Smart Electric Vehicle Battery Management System: This patent, filed by Anoop Arya, Manoj Arya, Amit Bhagat, Priyanka Paliwal, and Tripta Thakur, focuses on machine learning technology for developing a self-reconfigurable, flexible, and reliable model for battery management of electric vehicles. The invention provides a significant solution for battery management issues in electric vehicles based on integrating artificial intelligence with grid-connected vehicles.
A classical artificial intelligence algorithm – support vector regression (SVR) algorithm is utilized to establish a precise model of the battery in the cloud. Finally, battery degradation quantification is done based on the rain-flow cycle counting algorithm for dealing with issues related to the battery’s aging based on the battery’s model. - Enhanced Vehicle Safety
The patent US11383720B2 by Soungmin Im, Hyun Kim, Ilyong Lee, and Sangmin Lee is about a vehicle control method and an intelligent computing device for controlling a vehicle. The processor can detect the gaze response speed of the driver by projecting a virtual object to a Head-Up Display (HUD) when determining that the driver is exhausted. The processor outputs a secondary warning and controls the vehicle to be driven in accordance with the secondary warning when the gaze response speed of the driver is less than a predetermined reference value. This method can potentially enhance the safety of vehicle operation.
End Note
The predictive battery performance with research has opened new avenues for innovation. Future applications are vast, from designing advanced batteries to predictive maintenance and energy management. Machine learning could revolutionize the recycling process of batteries and play a key role in managing the power needs of the expanding Internet of Things.
The scope of future patents in predictive battery performance extends to predictive maintenance, energy management, and recycling processes. As algorithms evolve, patents may cover advanced battery designs, IoT power management, and innovative battery health diagnostics. This diversification indicates a broadening scope for innovation, with potential applications ranging from smart grid integration to personalized energy storage solutions.
The journey has just begun, and the future holds exciting possibilities in this dynamic field.