The state of charge SOC and state of health SOH of a battery are important indicators of battery endurance and lifetime. In order to solve the problem of time-varying battery parameters improve the accuracy of SOC estimation and reduce the hardware computation a joint multi-timescale online parameter identification algorithm with a doubleextended Kalman filter was proposed. The multi-timescale DEKF algorithm based on the second-order RC equivalent circuit model was used for the online parameter identification and SOC estimation of the 18 650 ternary lithium battery with the slow-varying characteristics of the battery parameters and the fast-varying characteristics of the battery state. Through the test verification of the Federal Urban Driving Program FUDS the average absolute errors of the SOC estimation of the multi-time scale DEKF algorithm and the traditional offline identification EKF algorithm were 0. 97% and 2. 46% respectively the rms errors were 1. 19% and 2. 69% and the maximum error of the capacity estimation to the reference value was only 0. 007 72 Ah. The experimental results show that the proposed time-scale DEKF algorithm has better robustness and SOC estimation accuracy and can respond to the SOH variation trend in real time.