SOH estimation method for lithium-ion battery packs under real-world operating conditions based on a new attenuated model without additional experiments

Estimating the health of state (SOH) of electric vehicle battery packs is key to ensuring their safety and reliability. However, the time-consuming degradation experiments to generate battery SOH labels hamper the deployment of relevant methods. In addition, due to the complexity of real-world operating conditions, accurate SOH estimation of battery packs in actual applications remains challenging. Therefore, in this paper, we first introduce a new attenuation SOH model based on constant current charge time (CCCT) to avoid the full-life tests of the battery for SOH labels' generation. The SOH labels can be generated online and updated. Thus, the challenge of obtaining SOH labels in real-world vehicle evaluation can be addressed. Secondly, to make the SOH estimation of battery packs under real-world operating conditions reasonable and reliable, a multi-dimensional feature extraction and fusion architecture is presented. Features are extracted from four dimensions. The battery operating parameters and user behaviour characteristics are comprehensively considered. Finally, we construct an SOH estimation model based on WOA-LightGBM architecture to improve the estimation accuracy by combining WOA's hyperparameter optimization with LightGBM's ability to capture complex data patterns. The experimental results show that the proposed method has high accuracy and robustness, the mean absolute error is only 0.074 %, R2 is 0.995, and the reliability of the SOH label defined by CCCT is verified.