Optical-sensing-oriented remaining useful life estimation of lithium-ion batteries using ALA-tuned VMD and a BiTCN-AM model

Qiushi Xie

School of Electronic Information and Communications
Huazhong University of Science and Technology, Wuhan, China

In Proc. SPIE, 2026.09

Code Dataset Paper · SPIE Method video BibTeX

Volume 14327 · Article 143271X · DOI: 10.1117/12.3122481

Adaptive signal decomposition meets bidirectional temporal learning for battery capacity prediction and remaining useful life estimation.

Method in motion

ALA–VMD → BiTCN–AM → RUL

From a capacity sequence to VMD components, bidirectional features, attention, and the reconstructed capacity trajectory.

Overview

The Artificial Lemming Algorithm (ALA) selects the mode number and penalty factor for Variational Mode Decomposition (VMD). A Bidirectional Temporal Convolutional Network with Attention Mechanism (BiTCN-AM) learns from the decomposed capacity signals. The model outputs are aggregated to reconstruct the capacity trajectory and estimate end of life at the failure threshold. Experiments in the paper use the NASA and CALCE battery datasets.

The method

01

Complete framework

Data preprocessing · Adaptive decomposition · Hybrid prediction · RUL estimation

02

Adaptive data decomposition

ALA selects (K, α); VMD decomposes the capacity sequence.

The NASA B5 decomposition contains eight IMFs. The first component and the sum of components 2–8 form the two prediction inputs.

03

Bidirectional temporal features and attention

Forward TCN · Reverse TCN · Feature fusion · Weighted aggregation

Paper architecture: the two branches form temporal features, followed by attention scores, weights, and a weighted output representation.

04

Capacity prediction and RUL estimation

Aggregated prediction · Capacity reconstruction · Failure threshold

NASA B5 · Test indices 61–167 · Capacity RMSE: 0.01137 Ah. The true and predicted trajectories first fall below 1.40 Ah at index 124.

BibTeX

Download .bib
@inproceedings{xie2026batteryrul,
  author = {Xie, Qiushi},
  title = {Optical-sensing-oriented remaining useful life estimation of lithium-ion batteries using ALA-tuned VMD and a BiTCN-AM model},
  booktitle = {Third International Conference on Electronics, Electrical, and Control System (EECS 2026)},
  publisher = {SPIE},
  volume = {14327},
  year = {2026},
  doi = {10.1117/12.3122481},
  url = {https://doi.org/10.1117/12.3122481}
}