State of Charge Estimation of Multi-Layer Structured Collaborative Lithium-Ion Batteries Considering Non-Gaussian Noise Interference

To address the issue of the reduction in the accuracy of state of charge (SOC) estimation of lithium-ion batteries caused by the interference of non-Gaussian colored noise in the input current, a multi-layer structure collaborative noise optimization method is proposed. Firstly, the adaptive noise integrated empirical mode decomposition (ICEEMDAN) is employed to process the input original signal, and the improved wavelet adaptive threshold function is utilized for noise reduction and reconstruction. Subsequently, the improved climber optimization algorithm (SCMTBO) is adopted to optimize the noise covariance matrix, and an adaptive extended Kalman filter algorithm under colored noise improvement (ACEKF) is put forward. A simulation model is constructed for testing in the Matlab/Simulink environment. The root mean square error (RMSE) of the ICEEMDAN - SCMTBO - ACEKF (IS-ACEKF) battery state estimation structure is 0.0022. In the robustness tests, under different working conditions, temperatures, colored noises, and initial SOC deviations, both the RMSE and the mean absolute error (MAE) remain at relatively low levels, which indicates that the proposed method can achieve high-precision SOC estimation and exhibits good robustness under non-Gaussian colored noise.