BatteryML currently supports public datasets that are applicable for battery lifetime prediction in early cycles. Please download the raw files from the data source and then run scripts/preprocess.py for data preparation.
MATR datasets contain four batches of lithium-ion phosphate (LFP)/graphite cells. Download the three batches used in [1] and the last batch used in [2]. Create the folder data/raw and place the four batches in data/raw/MATR:
data/raw/MATR
├── 2017-05-12_batchdata_updated_struct_errorcorrect.mat
├── 2017-06-30_batchdata_updated_struct_errorcorrect.mat
├── 2018-04-12_batchdata_updated_struct_errorcorrect.mat
└── 2019-01-24_batchdata_updated_struct_errorcorrect.mat
Finally, run the preprocessing script
python scripts/preprocess.pyHUST dataset contains 77 LFP cells using different discharge protocols [3]. The dataset is available via Mendeley Data. Create the folder data/raw and place raw file in data/raw/HUST:
data/raw/HUST
└── hust_data.zip
Finally, run the preprocessing script
python scripts/preprocess.pyCALCE dataset [4] is publicly available at here. Download the CS2 and CX2 series batteries and organize them as follows.
├── CS2_33.zip
├── CS2_34.zip
├── CS2_35.zip
├── CS2_36.zip
├── CS2_37.zip
├── CS2_38.zip
├── CX2_16.zip
├── CX2_33.zip
├── CX2_34.zip
├── CX2_35.zip
├── CX2_36.zip
├── CX2_37.zip
└── CX2_38.zip
Note that we did not use all the cells due to data format issues. We plan to include more cells in future. Run the preprocessing script to convert the files.
python scripts/preprocess.pyNote that the processing may take hours due to the many excel files.
RWTH dataset [5] is available at here.
data/raw/RWTH
└── raw.zip
Finally, run the preprocessing script
python scripts/preprocess.pySNL[6], UL-PUR[7], and HNEI [8] datasets are hosted by Battery Archive. They current no longer provide direct download links to these datasets. Users should apply for access first. After downloading the raw files, the file structure should look like
data/raw
├── HNEI
| ├── HNEI_18650_NMC_LCO_25C_0-100_0.5-1.5C_a_cycle_data.csv
| ├── HNEI_18650_NMC_LCO_25C_0-100_0.5-1.5C_a_timeseries.csv
| ├── HNEI_18650_NMC_LCO_25C_0-100_0.5-1.5C_b_cycle_data.csv
| ├── HNEI_18650_NMC_LCO_25C_0-100_0.5-1.5C_b_timeseries.csv
...
├── SNL
| ├── SNL_18650_LFP_15C_0-100_0.5-1C_a_cycle_data.csv
| ├── SNL_18650_LFP_15C_0-100_0.5-1C_a_timeseries.csv
| ├── SNL_18650_LFP_15C_0-100_0.5-1C_b_cycle_data.csv
| ├── SNL_18650_LFP_15C_0-100_0.5-1C_b_timeseries.csv
...
├── UL_PUR
| ├── UL-PUR_N10-EX9_18650_NCA_23C_0-100_0.5-0.5C_i_cycle_data.csv
| ├── UL-PUR_N10-EX9_18650_NCA_23C_0-100_0.5-0.5C_i_timeseries.csv
| ├── UL-PUR_N10-NA7_18650_NCA_23C_0-100_0.5-0.5C_g_cycle_data.csv
| ├── UL-PUR_N10-NA7_18650_NCA_23C_0-100_0.5-0.5C_g_timeseries.csv
...
Finally, run the preprocessing script
python scripts/preprocess.py[1] Severson et al. "Data-driven prediction of battery cycle life before capacity degradation." Nature Energy volume 4, pages 383–391 (2019).
[2] Attia et al. "Closed-loop optimization of fast-charging protocols for batteries with machine learning." Nature 578, 397–402 (2020).
[3] Ma et al. "Real-time personalized health status prediction of lithium-ion batteries using deep transfer learning." Energy & Environmental Science 15.10 (2022): 4083-4094.
[4] He et al. "Prognostics of lithium-ion batteries based on dempster–shafer theory and the bayesian monte carlo method." volume 196(23), pages 10314–10321, 2011.
[5] Li et al. "One-shot battery degradation trajectory prediction with deep learning." Journal of power sources, page 230024, 2021.
[6] Preger et al. "Degradation of commercial lithium-ion cells as a function of chemistry and cycling conditions." Journal of The Electrochemical Society, 167:120532, 2020.
[7] Juarez-Robles et al. "Degradation-safety analytics in lithium-ion cells: Part i. aging under charge/discharge cycling." Journal of The Electrochemical Society, 167:160510, 2020.
[8] Devie et al. "Intrinsic variability in the degradation of a batch of commercial 18650 lithium-ion cells." Energies, 11:1031, 2018.