Abstract:

Accurate estimation of the state-of-charge （SOC） with respect to the power in a lithium battery is the key to its safe and reliable use. Temperature significantly influences the usage of the lithium power battery. The advantages and disadvantages are compared by comprehensively analyzing the impact of various factors on the SOC for lithium batteries combined with various existing SOC estimation methods. Considering a temperature change from -10 ℃ to 40 ℃, a hybrid pulse power characteristic was obtained with respect to an AVIC 50 A·h lithium battery at intervals of 10 ℃. Based on the experimental data, the battery’s parameters were identified according to the least squares principle. The characteristics of the battery were explored with different temperature parameters, and the Thévenin equivalent circuit model （adapted for temperature changes） was established. The error variance matrix may gradually lose its positive definiteness or symmetry, resulting in filter divergence, because all the computer algorithm programs may be affected by word limitations or calculation errors. To solve this problem, the extended Kalman filter （EKF） algorithm is improved by square root decomposition of the EKF algorithm to accurately estimate the SOC. Through the simulation of the vehicle operating conditions of the China National Aviation’s ternary lithium battery, the simulation verification algorithm is used to estimate the effect under variable temperature conditions. The results reveal that the maximum error of SOC estimation based on the Thévenin equivalent circuit model at the considered variable temperature is less than 1.5% with an average error of 0.37%, which is better than that of the EKF algorithm. The improved EKF algorithm based on square root decomposition can rectify the error of the initial value of SOC and realize SOC estimation at different temperatures without depending on the accuracy of the initial value.

Key words: power Lithium-ion battery, SOC estimation, temperature, parameter identification, Thévenin model, improved extended Kalman filter