On the robustness of CNN-augmented sequential models for Li-ion battery RUL prediction under data scarcity

[1] The script "<code>CNN_ContiFormer_for_CALCE_Flexible.py"</code> corresponds to the "CNN-Transformer (Pre-LN)" model for the CALCE dataset, and its results are presented in Table 1 of the paper.<br>[2] The script "CNN_GRU_for_CALCE.py<code>"</code> corresponds to the "CNN_GRU" model for the CALCE dataset, and its results are presented in Table 1 of the paper.<br>[3] The script "CNN_LSTM_for_CALCE.py<code>"</code> corresponds to the "CNN-LSTM" model for the CALCE dataset, and its results are presented in Table 1 of the paper.[4] The script "CNN_ODE_RNN_for_CALCE_Flexible.py<code>"</code> corresponds to the "CNN-Neural ODE" model for the CALCE dataset, and its results are presented in Table 1 of the paper.[5] The script "CNN_Transformer_for_CALCE_Flexible.py<code>"</code> corresponds to the "CNN-Transformer" model for the CALCE dataset, and its results are presented in Table 1 of the paper.[6] The script "CNN_GRU.py<code>"</code> corresponds to the "CNN-GRU" model for the NASA dataset, and its results are presented in Table 1 of the paper.[7] The script "CNN_GRU_Mixed scarcity experiment.py<code>"</code> corresponds to the "CNN-GRU(Mixed scarcity experiment)" model for the NASA dataset, and its results are presented in Table 3 of the paper.[8]The script "CNN_LSTM.py<code>"</code> corresponds to the "CNN-LSTM" model for the NASA dataset, and its results are presented in Table 1 of the paper.[9]The script "CNN_ODE_RNN.py<code>"</code> corresponds to the "CNN-Neural ODE" model for the NASA dataset, and its results are presented in Table 1 of the paper.[10] The script "CNN_LSTM_mixed scarcity experiment.py<code>"</code> corresponds to the "CNN-LSTM(Mixed scarcity experiment)" model for the NASA dataset, and its results are presented in Table 3 of the paper.[11]The script "CNN_ODE_RNN_Mixed scarcity experiment.py<code>"</code> corresponds to the "CNN-Neural ODE(Mixed scarcity experiment)" model for the NASA dataset, and its results are presented in Table 3 of the paper.[12]The script "CNN_Transformer.py<code>"</code> corresponds to the "CNN-Transformer" model for the NASA dataset, and its results are presented in Table 1 of the paper.[13]The script "CNNContiFormer.py<code>"</code> corresponds to the "CNN-Transformer (Pre-LN)" model for the NASA dataset, and its results are presented in Table 1 of the paper.[14]The script "P-values were calculated (NASA).py<code>"</code> corresponds to different models for the NASA dataset, and its results are presented in Table 2 of the paper. This code needs to replace the data path.[15]The script "ICA Feature Extraction.py<code>"and "</code>Run Physics-Based Experiments.py<code>"</code> corresponds to "ICA-GRU" model for the NASA dataset, and its results are presented in Table 4 of the paper. [16]The script "calculate_and_visualize_gradcam.py<code>"and "</code>visualize_cross_cycle_gradcam.py<code>" are</code> the map visualization of cross-cycle Grad-CAM saliency (B0005), and this figure is presented in the supporting information files of the paper.