Case studies: AI solutions for battery validation 

The shift to electric vehicles is driving disruption throughout the automotive industry. 

Battery technology is a primary driver of this disruption. Engineers are pushing to design, develop, and produce new batteries that last longer, charge faster, and deliver more power to meet market needs. Because battery simulation is difficult, battery cell manufacturers and OEMs are forced to rely more heavily on physical testing of battery technologies to validate their performance, reliability, and safety. 

Artificial intelligence (AI) and machine learning are attracting interest from battery engineers to better deal with the intractable physics of battery design and development. Every day, more research is published proving that machine learning can address many applications across the battery development workflow. 

This paper highlights specific challenges where battery engineers are applying machine learning to automate, accelerate, or reduce various aspects of battery design, testing, and development. We highlight specific applications and the Monolith tools being applied to address it. With Monolith, battery development teams are discovering that AI and machine learning is a reality today, able to help them define smarter test plans, find errors in your data and speed time to market throughout the engineering workflow. 

 Some of the case studies highlighted include: 

• Cell design: Reduce cathode design by 50% with AI 
• Cell design: Reduce validation burden to accelerate new battery technology 
• Test data validation: Automate data segmentation to find 90% of known errors 
• Test data validation: AI-guided data segmentation enables cell design optimisation 
• Test plan optimisation: Reduce battery aging matrix by 58% 
• Test plan optimisation: Reduce aging tests by 40% and cell repetitions by 75%
• Test plan optimisation: Reduce capacity fade matrix by 50% 
• System calibration: Reduce physical testing required for ECU calibration