State estimation and visualization of lithium‑ion battery using transformer Autoencoder model

Tae-Geol Woo; Beom-Jun Kim; In-Ho Cho; Kang-Moon Park

State estimation and visualization of lithium‑ion battery using transformer Autoencoder model

Vol. 25, No. 3, pp. 500-509, Mar. 2025

10.1007/s43236-025-00995-6

State of Health (SOH)

Abstract

This study developed and evaluated artificial neural network models for estimating and visualizing the State of Health (SOH) of lithium-ion batteries (LIBs) used in railway vehicles. The models, which include Autoencoder, Long Short-Term Memory (LSTM)-Autoencoder, Attention LSTM-Autoencoder, and Transformer Autoencoder, were trained on large-scale time-series data of LIB voltage, current, and temperature. The Transformer Autoencoder model demonstrated substantial performance improvements, achieving over a 99% enhancement compared to the basic Autoencoder model and up to 82% improvement over the Attention LSTM-Autoencoder model. Unlike classical Transformer models, which typically focus on compressing data into high-dimensional spaces, the Autoencoder approach is applied to the Transformer model, facilitating low-dimensional compression and clustering of the data. This clustering technique was further employed to visualize battery health by converting the clustered data into RGB values, offering an intuitive representation of the SOH. By overcoming the limitations of traditional methods, this novel approach provides an eff ective means of assessing battery condition. These findings indicate that the proposed method could notably enhance battery management systems, leading to a safe and reliable operation of electric mobility systems.

Statistics

Show / Hide Statistics

Cumulative Counts from September 30th, 2019
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.

Cite this article

[IEEE Style]

T. Woo, B. Kim, I. Cho, K. Park, "State estimation and visualization of lithium‑ion battery using transformer Autoencoder model," Journal of Power Electronics, vol. 25, no. 3, pp. 500-509, 2025. DOI: 10.1007/s43236-025-00995-6.

[ACM Style]

Tae-Geol Woo, Beom-Jun Kim, In-Ho Cho, and Kang-Moon Park. 2025. State estimation and visualization of lithium‑ion battery using transformer Autoencoder model. Journal of Power Electronics, 25, 3, (2025), 500-509. DOI: 10.1007/s43236-025-00995-6.