Multivariate Lithium-ion Battery State Prediction with Channel-Independent Informer and Particle Filter for Battery Digital Twin

Accurate State-of-Health (SOH) prediction is critical for the safe and efficient operation of lithium-ion batteries (LiBs). However, conventional methods struggle with the highly nonlinear electrochemical dynamics and declining accuracy over long-horizon forecasting. To address these limitations, this study proposes CIPF-Informer, a novel digital twin framework that integrates the Informer architecture with Channel Independence (CI) and a Particle Filter (PF). The CI mechanism enhances robustness by decoupling multivariate state dependencies, while the PF captures the complex stochastic variations missed by purely deterministic models. The proposed framework was evaluated using the Massachusetts Institute of Technology (MIT) battery dataset against benchmark deep learning models. Results demonstrate that CIPF-Informer consistently achieves superior performance, in multivariate and long sequence forecasting scenarios. By effectively synergizing a model-based method with a data-driven model, CIPF-Informer provides a more reliable pathway for advancing Battery Management System (BMS) technologies, contributing to the development of safer and more sustainable energy storage systems.