Novel platform generates massive battery fault data, revolutionizing EV safety research

A groundbreaking data generation platform developed by researchers at RWTH Aachen University's Center for Ageing, Reliability and Lifetime Prediction of Electrochemical and Power Electronic Systems (CARL) promises to transform how battery faults are detected and diagnosed in electric vehicles, potentially saving manufacturers millions in testing costs while enhancing safety.

As electric vehicles become increasingly mainstream, ensuring the safety and reliability of lithium-ion battery packs remains a paramount concern for manufacturers and consumers alike. Battery failures, though rare, can have catastrophic consequences and significantly impact market confidence in EV technology.

"Data-driven fault diagnosis methods offer tremendous potential for early detection of battery issues, but they require massive amounts of real-world data that is prohibitively expensive and time-consuming to collect through traditional laboratory testing," explains the research team. "Our platform addresses this fundamental challenge by generating realistic battery pack data that includes various fault scenarios."

The innovative platform combines electrical, thermal, and aging models with sophisticated fault simulations to generate data for 8,000 virtual EV battery packs. What sets this research apart is its comprehensive approach to modeling real-world variations that affect battery performance. Moreover, the researchers successfully modeled four critical fault types: hard short circuits, soft short circuits, abnormal internal resistance, and abnormal contact resistance—all potentially catastrophic issues that are difficult to study in real-world settings due to their dangerous nature.

The technology offers immediate benefits for battery management system developers who can now train advanced fault detection algorithms using abundant, diverse, and accurately labeled data. For EV manufacturers, this means faster development cycles and potentially earlier detection of battery issues before they become dangerous.

Looking ahead, the platform could be expanded to include additional fault types and battery chemistries as the EV market evolves. The researchers suggest that cloud-based implementations could allow for even larger data generation capabilities, potentially creating industry-wide databases for battery fault patterns. The team is also exploring how this technology could be integrated with real-time monitoring systems in vehicles, creating hybrid diagnostic approaches that combine simulation-trained algorithms with actual vehicle data.

By accelerating the development of more sophisticated battery fault detection systems, this research represents a significant step toward improving the safety, reliability, and consumer acceptance of electric vehicles—ultimately supporting the global transition to sustainable transportation.

Reference

Author:

Daniel Luder ^a b c, PraiseThomas John ^a b c, Paul Busch ^a b c, Martin Börner ^a b c, Wenjiong Cao ^a b c, Philipp Dechent ^a b c, Elias Barbers ^a b c e, Stephan Bihn ^a b c, Lishuo Liu ^a b c d, Xuning Feng ^d, Dirk Uwe Sauer ^a b c e, Weihan Li ^a b c

Title of original paper: Big data generation platform for battery faults under real-world variances

Article link:  https://www.sciencedirect.com/science/article/pii/S2773153725000325

Journal:  Green Energy and Intelligent Transportation

DOI: 10.1016/j.geits.2025.100282

Affiliations:

^a Center for Ageing, Reliability and Lifetime Prediction of Electrochemical and Power Electronic Systems (CARL), RWTH Aachen University, Campus-Boulevard 89, Aachen, Germany

^b Institute for Power Electronics and Electrical Drives (ISEA), RWTH Aachen University, Campus-Boulevard 89, Aachen, Germany

^c Juelich Aachen Research Alliance, JARA-Energy, Germany

^d State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing 100084, China

^e Helmholtz Institute Münster (HI MS), IMD-4, Forschungszentrum Jülich, Germany