Abstract: Since the SOC (state of charge) of a lithium-ion battery cannot be directly measured, and can only be estimated by the external output characteristics of the battery. Taking lithium-ion phosphate lithium battery as the research object and considering the various complex nonlinear characteristics of the battery, the electrochemical impedance characteristics of the battery were analyzed. To improve the traditional equivalent circuit model, the fractional order equivalent circuit model was established using the constant phase element (CPE) combined Genetic Algorithm and hybrid impulse dynamic test to identify the parameters of the fractional equivalent circuit model offline; Based on the Extended Kalman Filter algorithm, the lithium battery SOC estimation model was built by a Fractional Order Extended Kalman Filter (FEKF) algorithm; According to the DST (dynamic stress test) conditions, charging and discharging plan for the lithium battery were designed, and the battery current and voltage data were collected in real time at an ambient temperature of 25 ℃, which was input into the model established by Matlab to estimate the SOC of the battery. Compared with the simulated SOC of the traditional second-order Thevenin circuit model, the SOC estimated based on the FEKF algorithm has higher accuracy and smaller volatility, and the error is less than 0.72% and the RMSE is only 0.24%.