Machine learning is a subset of artificial intelligence, which refers to the process of computers drawing conclusions from a dataset, which can be a powerful tool across various sectors, including the battery market. IDTechEx’s report, “AI-Driven Battery Technology 2025-2035: Technology, Innovation and Opportunities“, demonstrates the use of machine learning in disrupting stages of a battery’s lifecycle, to provide increased efficiency, safety, and reduce costs.
Challenges within the battery market
Lithium-ion batteries are close to their ceiling of energy density, with new chemistries needing to be discovered and developed faster to keep up with increased energy demand. Liquid electrolytes are also proving to be a challenge, as it is difficult to make them safe for applications under high temperatures and pressures, and can therefore be a safety concern in electric vehicle applications. Faster cell testing methods, in-life diagnostics, and monitoring carried out by machine learning could help manage these potential risks.
Material extraction is another hurdle, with lithium-ion batteries relying on sourcing critical materials such as lithium, nickel, cobalt, and copper. Extracting these materials is costly and creates environmental concerns with high carbon emissions. There is a limited supply, meaning developments to obtain a higher production yield are needed, as well as battery repurposing and recycling to limit the use of rare metals.
AI implementation within the battery lifecycle
Machine learning methods can be used at various stages of a battery’s lifecycle, from the raw material extraction all the way down to the recycling and refurbishment stage at the very end.
De novo designs and virtual screening could allow for the use of cheaper and more readily available materials within batteries, according to IDTechEx. Material property screening is a process for which AI is necessary, as the advantages cannot easily be replicated by other methods.
Manufacturing parameter optimization and root-cause optimization can be used to optimize component design and cell structure for higher energy density and safety, while also increasing yield and quality of batteries. Cell simulation can then be used to reduce cycling time and increase yield. Simulation in factory acceptance testing (FAT) is a machine learning approach to detect defective battery modules and the root causes of structural defects before they are shipped, preventing extra costs down the line.
As the batteries are in-life, data-based diagnostics can accurately keep track of their state of health to manage any safety concerns and deal with them promptly and effectively to prevent further damage. Lastly, second-life diagnostics can determine the potential of a battery for new applications, identifying the volume and quality of reusable cells and packs, and where they would perform best in their second life.
Benchmarking
IDTechEx’s report, “AI-Driven Battery Technology 2025-2035: Technology, Innovation and Opportunities”, benchmarks the readiness and necessity levels of machine learning approaches within the battery market. The report evaluates AI necessity, current value proposition, theoretical value proposition, development challenges, and costs.
Material property screening, despite having a weak current value proposition due to the technology being immature, is predicted in the future to be an essential part of cell development. Manufacturing process optimization, state of health calculation, and charge-discharge profile optimization are all relatively mature technologies however, the specific advantages provided by AI are not as significant as for other uses.
For more information on this report, including downloadable sample pages, please see www.IDTechEx.com/AIBattTech.
For the full portfolio of battery and energy storage market research available from IDTechEx, please visit www.IDTechEx.com/Research/ES.
