Researchers have developed a hybrid modeling framework that enables accurate, real-time prediction of temperature distribution in lithium-ion battery cells used in electric vehicles (EVs), addressing a key challenge in battery thermal management.
Published in Results in Engineering, the study presents a combined finite element method (FEM) and neural network (NN) approach designed to overcome the computational limitations of traditional physics-based thermal models.
Hybrid FEM–neural network workflow for predicting mean, maximum, and minimum temperatures in EV battery cells under varying loads and ambient conditions. (Image: ScienceDirect)
The FEM model was first calibrated against experimental data, then used to generate more than 42,000 thermal data samples representing simultaneous variations in charge–discharge cycles and ambient temperatures between 20° and 40° C, conditions representative of real-world EV operation.
Using this dataset, multiple neural-network architectures were evaluated. A model with two hidden layers and 512 neurons achieved the best balance of accuracy and efficiency, reaching coefficient-of-determination values up to 0.98 and a root-mean-square error of 0.58° C.
While deeper or simpler models either increased computation time or reduced accuracy, the optimized model reduced computational demand by approximately 94% compared with standalone FEM simulations. Training time was approximately 2.5 minutes, with inference time measured at 0.002 seconds.
Accurate and fast temperature estimation is critical for electric-vehicle battery systems, where thermal conditions directly affect performance, fast-charging capability, degradation rate, and safety margins. By enabling high-fidelity temperature prediction at speeds suitable for real-time control, the proposed framework demonstrates a practical pathway for integration into advanced battery management systems.
The researchers note that future work will focus on incorporating realistic drive-cycle profiles, battery aging effects, and module-level thermal validation to further align the approach with production EV battery systems.
The study was conducted by Arsanchai Sukkuea, Komsan Srivisut, and Padej Pao-la-or from the School of Electrical and Computer Engineering at Suranaree University of Technology in Thailand and was published in Results in Engineering in 2025.
Filed Under: Batteries, Technology News, Thermal Management