Fig. 3: The model performance in terms of mean absolute percentage error (MAPE) and energy-environmental benefit. | Nature Communications

Fig. 3: The model performance in terms of mean absolute percentage error (MAPE) and energy-environmental benefit.

From: Generative learning assisted state-of-health estimation for sustainable battery recycling with random retirement conditions

Fig. 3

a The parity plot of the model training results (reconstruction), with feature U1 at the selected state of charge (SOC) condition (35% and 50%), is illustrated. b The heatmap of the model training performance for features U1 to U21. c SOC simulation of the retired batteries under random retirement scenarios, including interpolation and extrapolation cases. d The data generation performance under the random retirement scenarios, where error bars indicate the standard deviation (\({{{\rm{\sigma }}}}\)) computed across 21-dimensional features (n = 21 at each SOC in each case) and the height of the bar indicates its mean. e The electricity cost savings and CO2 emission reductions for different battery retirement scales using the different data generation case settings. Note that the battery retirement scale is logarithmic. All results are from NMC, 2.1 Ah batteries for illustration. Source data are provided as a Source Data file.

Back to article page