Enhanced CNN-LSTM attention model with adaptive loss for lithium-Ion battery state of health estimation

The data folder contains both the datasets and the experimental results. The original dataset was sourced from NASA and expanded using several data augmentation techniques. The dataset files are stored under the NASA subdirectory. During the experimental process, the original dataset and the three augmented versions—generated through slicing, flipping, and linear mixup—were each placed into the total folder, resulting in four separate experimental results. All augmented datasets were created using custom data augmentation scripts. The experimental results include evaluations of the baseline function, the improved main function, hyperparameter optimization, loss function comparisons, and ablation study outcomes.本数据文件夹同时包含数据集与实验结果。原始数据集源自美国国家航空航天局（NASA），并通过多种数据增强技术完成扩充。数据集文件存储于NASA子目录下。实验过程中，原始数据集与通过切片、翻转及线性混合（linear mixup）生成的三类增强数据集均被置入总文件夹，最终得到四组独立的实验结果。所有增强数据集均通过自研数据增强脚本生成。 本次实验结果涵盖基线函数、改进后主函数的性能评估、超参数优化、损失函数对比以及消融实验（ablation study）的相关结果。