Objective Accurate estimation of the state of charge SOC of lithium batteries is of great significance in the field of new energy vehicles. Therefore a dual unscented Kalman filter algorithm FOUKF-FOSUKF is proposed which combines the fractional-order spherical unscented Kalman filter FOUKF and the fractional-order unscented Kalman filter FOSUKF to estimate the SOC of batteries. Methods Firstly the parameters of the battery model are identified offline using the adaptive genetic algorithm AGA . Then the fractional-order unscented Kalman filter FOUKF is employed for online parameter identification to estimate and update the parameters in the fractional-order model of the lithium battery in real-time. Finally the proposed combined algorithm FOUKF-FOSUKF is used to estimate the state of charge SOC of the lithium battery. The accuracy of the proposed method is verified and compared with the traditional integer-order spherical unscented Kalman filter SUKF and the fractional-order spherical Kalman filter FOSUKF under two conditions dynamic stress test DST and US06. Results During the SOC estimation process the SOC error and voltage error of FOUKF-FOSUKF are significantly lower than those of the traditional SUKF and FOSUKF. This algorithm can effectively estimate the parameters in the battery model reduce the error in terminal voltage estimation and improve the accuracy of SOC estimation. Conclusion FOUKF-FOSUKF has higher accuracy smaller error stronger applicability and better convergence compared with the SUKF and FOSUKF algorithms in estimating the SOC of lithium batteries.