Abstract: The state of charge(SOC) of lithium-ion batteries is a key indicator closely related to battery performance and safety in energy storage systems, thus requiring accurate estimation. Based on the second-order resistance capacitance(RC) equivalent circuit model, a bias compensation variable forgetting factor recurrent least square(BC-VFFRLS) is proposed for parameter identification of dynamically changing model parameters. The multi innovation cubature kalman filtering(MSCKF) algorithm is used to achieve SOC estimation for lithium batteries, thereby improving the accuracy of SOC estimation. Verify parameter identification through HPPC operating conditions, and verify SOC estimation results through BBDST and DST operating conditions. The verification results show that the SOC error under BBDST working condition is stably controlled within 0.72%, and the SOC error under DST working condition is stably controlled within 1.02%. The experimental results fully verify that the proposed algorithm has good accuracy and convergence.