Future battery development will come from AI

The transition to an all-electric future depends on low-cost, high-performance and safer batteries. Efforts to optimize the energy density and power of batteries with next-generation electrochemical compositions, such as solid-state, have achieved varying degrees of success. However, none of them have yet reached the commercialization stage to meet the explosive demand for advanced technologies such as electric vehicles, medical devices, drones and energy storage solutions.

As China, Europe, and the United States compete to dominate the global battery market—expected to reach $279.7 billion by 2027—AI has emerged as a very promising tool for accelerating the pace of innovation.

To achieve this goal, however, next-generation batteries must be able to charge quickly and safely. They also need to exceed current performance standards while keeping weight low and battery materials easy to mass-produce.

Researchers have spent decades working on solutions, but progress has been hampered by slow experiments, long cycles, and a difficult discovery process. AI could help solve these long-standing challenges and shorten the process of evaluating battery materials, battery structures, and chemistries from years to months.

Shorten lengthy evaluation periods

The traditional way to get data on battery performance involves pumping energy into cells until they die. Researchers may have to spend years putting batteries through the charge and discharge cycle thousands of times to get the results they need. This is how researchers predict battery degradation, a process that is critical to developing safer, nonflammable batteries. Given the dramatic increase in demand for electric vehicles and residential solar storage, there's clearly no time to waste.

Taking a systems-level approach, battery scientists can apply AI to more effectively test and understand battery packs, their integration, and expected performance. This application of AI also includes various battery types, their different chemistries and expected performance, and can help determine the best way to distribute energy within multiple cells or battery packs.

In the past, researchers faced the daunting task of narrowing down the necessary alternative materials for next-generation battery applications. The process required evaluators to extract insights from the vast amounts of data collected from testing. Researchers could only operate at the speed at which machines could compute information, often taking years to make progress.

AI can unlock some useful material combinations that would not otherwise be considered. AI has produced interesting results in the field of discovering new materials, such as superconductors, and it also seems promising in the field of batteries.

Optimizing battery structure with AI

While most efforts to improve batteries over the past few decades have focused on electrochemistry, changing the physical properties of batteries has been shown to improve battery density, capacity, safety, and other key performance indicators. AI can help battery scientists better understand the relationship between structure and performance at the electrode level in order to design the best battery structure for any specific application. Depending on how the battery is used and other technical specifications, AI can make insightful suggestions on possible structural designs to optimize battery performance.

AI algorithms can even suggest possibilities based on emerging technologies and chemistries that haven’t been applied yet. This is like having a robust battery prototype factory, with huge benefits for the entire value chain in terms of time and cost savings.

For example, from an automotive perspective, a large part of the performance of electric vehicles is based on the battery cell. That's why it is so important to connect it with the capabilities of AI to better understand how to improve the BMS and improve the capabilities of the battery. This will help lay the foundation for the development of the next generation of power batteries.

Will AI determine who wins and who loses the battery race?

Although AI is still an emerging application in battery science, we are beginning to see its huge potential. For example, researchers from Stanford University, MIT, and Toyota Research Institute used AI to determine the best way to charge electric vehicle batteries in just 10 minutes. Traditional methods would require a 500-day evaluation process, but the team was able to use a highly targeted AI algorithm to determine the best charging method among 224 candidate solutions in just 16 days.

It's not just researchers, but also big companies. Volkswagen, in collaboration with Google, is using AI and quantum computing to simulate and optimize the structure of high-performance batteries. Panasonic claims that thanks to the help of AI, the number of charge and discharge cycles is greatly reduced when testing new designs. Although these are just a few examples, as machine learning technology develops, its applications and advantages will continue to increase.

The battery industry is an increasingly crowded race, with new and existing players all seeking the next great battery technology breakthrough. As the commercialization of the next generation of batteries will take 5-15 years, maintaining a competitive advantage may require the successful adoption of AI to accelerate the testing phase and identify areas for improved cost efficiency and performance.

[Reference article]

Next-Generation Batteries Will Be Brought to You by AI — Dr. Moshiel Biton