A Data-Driven, Sensor Placement Optimized Method for Intelligent Thermal Fault Prediction in Air-Cooled Cylindrical Li-Ion Battery Modules

View at publisher Summarise with VIRAT
Authors: Rojo Kurian Daniels, Aneesh Prabhakar
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Thermal runaway (TR) in lithium-ion batteries (LiBs) represents a significant safety concern in electric vehicles, requiring rapid and accurate fault localization to mitigate propagation risks. This study investigates TR fault localization in air-cooled cylindrical LiB modules by introducing and evaluating two randomized single-plane temperature sensor layouts to address the limited predictive accuracy of uniform single-plane distributions studied earlier. Temperature data for multiple TR fault positions under laminar and turbulent flow conditions are generated using a computational fluid dynamics model that has been validated with in-house experiments. A Pearson correlation coefficient (PCC) based feature selection method is applied to identify optimized sensor layouts, whose temperature data are then used to train a Random Forest (RF) classifier. The proposed approach achieves up to 100% accuracy in laminar and 94% in turbulent regimes, when validated against 100 external test cases. Results indicate that a structured single-plane layout, with sensors positioned between four neighboring cells, significantly improves prediction performance over random scattering, while multi-plane arrangements remain superior for overall accuracy. Furthermore, the results give EV manufacturers useful information for developing safer, more intelligent battery monitoring systems that may be used in commercial vehicles.

Fields of Study & Subjects

Battery Modeling and Management Advancements in Battery Materials Fault Detection and Control Systems

Publication details

Journal / source
IEEE Transactions on Industry Applications
Pages
1–11
Publication date
1 January 2026
Language
EN