Estimation of State of Charge (SOC) of the battery is an important part of battery design, and accurate power estimation is important to improve the efficiency of power equipment and the full utilization of batteries. In order to improve the accuracy and stability of Support Vector Regression (SVR) in SOC estimation, taking Improved Grey Wolf Algorithm (IGWO) to optimize SVR hyperparameters as an example, a process of finding the optimal SVR hyperparameter interval is established with the help of MATLAB automatic optimization, grid search and Occam's razor principle. Finally, after applying the IGWO-SVR model obtained from the optimized search interval to the National Aeronautics and Space Administration (NASA) Li-ion battery dataset for SOC prediction, the results show that the average error of the model after the optimized search interval is 1% on the test set, and the maximum error under different operating conditions is also minimized, which proves the good performance of the model in practical applications.