Lithium-Ion Battery State of Health Estimation Based on Time-Frequency Image Fusion and KAN

This paper proposes a method for estimating the SOH of lithium-ion batteries based on improved time-frequency image fusion and Kolmogorov-Arnold network (KANs). First, the time-frequency feature extraction process is simplified by employing weighted average image fusion and one-dimensional unfolded Euclidean distance calculation. Second, for the regression estimation task, The KANs are optimized using tensor operations and regularization simplification. Experiments conducted on the CALCE datasets demonstrate that the KANs model outperforms other comparative models in terms of estimation accuracy, computational efficiency, and generalization ability. On the CALCE dataset, the lowest root mean square error (RMSE) and mean absolute error (MAE) are 0.53% and 0.35%, respectively, while the highest coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) reaches 0.9962., confirming the effectiveness and superiority of the proposed method.