Lithium-Ion Battery Health State Prediction Based on VMD and DBO-SVR

Accurate estimation of the state-of-health (SOH) lithium-ion batteries is a crucial reference for energy management battery packs electric vehicles. It great significance in ensuring safe and reliable operation while reducing maintenance costs system. To eliminate nonlinear effects caused by factors such as capacity regeneration on SOH sequence improve prediction accuracy stability SOH, model based Variational Modal Decomposition (VMD) Dung Beetle Optimization -Support Vector Regression (DBO-SVR) proposed. Firstly, VMD algorithm used to decompose into series stationary mode components. Then, each component treated separate subsequence modeled predicted directly using SVR. address problem difficult parameter selection SVR, DBO optimize parameters SVR before training. Finally, values are added reconstructed obtain final prediction. In order verify effectiveness proposed method, VMD-DBO-SVR was compared with Empirical Mode Decomposition-Support (EMD-SVR), VMD-SVR methods NASA dataset. Experimental results show that has higher fitting degree, errors all within 1% better robustness.