Accurate and online capacity estimation is of extreme importance to maintain the continuous operation of lithium-ion batteries. This paper proposes an indirect capacity estimation method based on the incremental capacity features and model interpretability. First, the current and voltage data of the battery are collected in real-time to construct the incremental capacity curves. The dual filter, which consists of a moving average and a Gaussian filter, is then used to smooth the curves. To achieve satisfying filtering effects, the filter window size and calculation frequency are determined by comparing different sets of conditions. 15 multiple alternative features related to the curve peak position and area are extracted. The Shapley value method based on cooperative game theory is introduced to reduce the dimension of the feature vector and to determine the key features. Finally, an XGBoost model is established to map the relationship between three key features and the battery capacity. The contribution of each feature to the estimation is quantitatively analyzed based on its Shapley value. The proposed battery capacity estimation method is validated on the experimental datasets. The results show that the relative error of 94.7% of the test samples is less than 2%, and the overall average relative error is 0.877.
Online Capacity Estimation of Lithium-ion Battery Based on Incremental Capacity Analysis with Interpretable Features
Sae Technical Papers
SAE 2021 Intelligent and Connected Vehicles Symposium Part I ; 2021
2021-12-15
Conference paper
English
British Library Conference Proceedings | 2021
|