Noise-immune state of charge estimation for lithium-ion batteries based on optimized dynamic model and improved adaptive unscented Kalman filter under wide temperature range

Due to diverse vehicle driving conditions, it is difficult to ensure that lithium-ion batteries operate at a fixed ambient temperature. Meanwhile, the environmental noise accompanying vehicle driving can also affect the data sampling accuracy of batteries. To achieve a high-robustness state of charge (SOC) estimation for lithium-ion batteries at different ambient temperatures with noise effects, a noise-immune SOC estimation method under a wide ambient temperature range is proposed. First, based on the moving window principle, a high-precision fitting model of the open-circuit voltage is established. Second, based on the Multi-Verse Optimizer, a new dynamic model parameter identification method is proposed, while the complete optimized dynamic battery model is established relying on the data of Dynamic Stress Test at different temperatures. Third, to enhance the SOC estimation accuracy and stability, based on adaptive theory and matrix diagonalization theory, the un-scented Kalman filter is improved. Finally, the effectiveness and robustness of the proposed method are validated under two other working conditions with random noise added at various temperatures. Under every set of working conditions at all temperatures, the SOC estimation results can maintain stability after converging to the reference SOC, while root mean square errors and mean absolute errors under all cases do not exceed 1.5 %.

Automated summary

To achieve a high-robustness state of charge (SOC) estimation for lithium-ion batteries under different ambient temperatures with noise effects, a noise-immune SOC estimation method is proposed. The method includes establishing a high-precision fitting model of the open-circuit voltage based on the moving window principle, developing a new dynamic model parameter identification method using the Multi-Verse Optimizer, and improving the unscented Kalman filter based on adaptive theory and matrix diagonalization theory. The proposed method is validated to maintain stability and accuracy in SOC estimation at different temperatures and under varying working conditions.