Rapid Model-Free State of Health Estimation for End-Of-First-Life Electric Vehicle Batteries Using Impedance Spectroscopy

The continually expanding number of electric vehicles in circulation presents challenges terms end-of-life disposal, driving interest the reuse batteries for second-life applications. A key aspect battery is quantification relative condition or state health (SoH), to inform subsequent application and match similar capacity. Impedance spectroscopy has demonstrated potential estimation health, however, there difficulty interpreting results estimate reliably. This study proposes a model-free, convolutional-neural-network-based scheme high-power lithium-ion based on dataset impedance measurements from 13 end-of-first-life Nissan Leaf 2011 modules. As baseline, this compared with our previous approach, where parameters Randles equivalent circuit model (ECM) without dataset-specific adaptations ECM were extracted train deep neural network refined using Bayesian hyperparameter optimisation. It that small 128 samples, proposed method achieves good discrimination high low superior prediction accuracy model-based approach by RMS error (1.974 SoH%) peak (4.935 metrics improve fit quality. accomplished while maintaining competitive performance when previously SoH schemes.