Online state estimation of lithium-ion batteries using nonlinear ultrasonics with LSTM-based deep learning model

Online state estimation of lithium-ion batteries using nonlinear ultrasonics with LSTM-based deep learning model

eTransportation

10.1016/j.etran.2026.100613

https://doi.org/10.1016/j.etran.2026.100613

Santhakumar Sampath, Tam Van Huynh, Xuesong Yin, Zi Wen Tham, Andrew Ngo, Hoon Sohn, Lei Zhang

Lithium-ion batteries, Nonlinear ultrasonics, State-of-charge, Long short-term memory

18 June 2026

Sampath, S., Huynh, T. V., Yin, X., Tham, Z. W., Ngo, A., Sohn, H., & Zhang, L. (2026). Online state estimation of lithium-ion batteries using nonlinear ultrasonics with LSTM-based deep learning model. eTransportation, 29, 100613. https://doi.org/10.1016/j.etran.2026.100613

Accurate real-time state estimation is still one of the crucial issues for ensuring the safety and reliable operation of lithium-ion batteries. Current ultrasonic methods mainly rely on linear features such as time-of-flight. These features often lead to large estimation errors because the response depends mainly on second-order elastic constants. This study introduces a nonlinear ultrasonic method that integrates bispectral analysis and a long short-term memory (LSTM) network for real-time battery state estimation. A theoretical relationship between the nonlinear ultrasonic parameter and battery state is established based on the elasticity theory. Ultrasonic signals are acquired using a through-transmission configuration, and ultrasonic nonlinearity parameter is calculated from the reconstructed signals. The proposed method reduces the state of charge (SOC) estimation error to below 3% and improves estimation accuracy by 3.06 times compared to linear features. Under different C-rates (C/2, 1C and 2C), the root mean squared error for SOC estimation is 3.1%, 2.4% and 2.9%, respectively. The results demonstrate that nonlinear ultrasonics combined with deep learning improves real-time battery state estimation accuracy.

Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)

This research / project is supported by the A*STAR - Manufacturing, Trade, and Connectivity Young Individual Research Grants
Grant Reference no. : M24N8c0101

This research / project is supported by the Energy Market Agency - 2nd Energy Storage Grant
Grant Reference no. : EP014-ESGC2-0001

This research / project is supported by the A*STAR - Battery Remanufacturing for Improved Circular Ecosystems (BRICE)
Grant Reference no. : M24N2a0076

https://oar.a-star.edu.sg/communities-collections/articles/23442

2590-1168

Institute of Materials Research and Engineering