Lithium batteries have the advantages of high energy density and long cycle life, and are widely used in electric vehicle power plants. However, the operating conditions of vehicles are complex and variable, and the battery exhibits highly nonlinear properties, making it difficult to accurately calculate the state of charge (SOC) of the battery. In order to optimize the SOC estimation accuracy of lithium batteries, a fractional second-order RC model combined with Warburg elements was constructed, and a adaptive genetic algorithm was used for parameter identification. Combining multi innovation theory and extended Kalman filter filter algorithm, an ion battery SOC estimation algorithm based on multi innovation extended Kalman filter (MIEKF) was proposed, and the effectiveness of this method was verified by experimental data, which provided a new approach and practical support for improving the SOC estimation accuracy and the cycle life of vehicle mounted lithium batteries.