EV Battery Life Prediction: 4.6X Model Performance Boost

  • Battery
  • Predictive Analytics

Challenge

As electric vehicles (EVs) become increasingly commercialized, the demand for precise prediction of lithium-ion battery (Li-ion battery) lifespan has intensified. Efficient battery lifecycle management is challenged by the scarcity of comprehensive data, given that few EVs have been driven to the extent of their battery's warranty life. Traditional battery life predictions, which rely on basic criteria like vehicle and battery types, have proven inaccurate. This has heightened the need for a more personalized approach to battery life prediction, one that takes into account real-world driving patterns.

Approach

Data collection devices were attached to over 1,000 on-road EVs to gather real-time information on driving, charging, and parking behaviors. Through the analysis of battery management system data and driving patterns, a model was developed to predict the remaining battery life of commercial EVs. This model was further refined by weekly analysis of driving patterns, adapting to various driving conditions and environments to enhance its predictive accuracy.

Value Delivered

The AI models were successfully scaled to over 2,700 vehicles, leading to a significant improvement in the generation of driving scenarios – approximately twice as accurate as previous models. The weekly degradation models saw an even greater enhancement, becoming 4.6 times more accurate. This capability to monitor battery lifecycles has enabled more efficient operation of electric vehicles, allowing for timely battery replacement and informed decisions on battery reuse based on their residual lifespan.

Want to learn more about this use case?
Get in touch with our industrial AI experts.


Talk to an AI Expert