In order to achieve the optimal adhesion utilization of high-speed trains, this paper designs the optimal creep speed search method. Firstly, the wheel speed and motor torque are used to estimate the current adhesion coefficient; secondly, the estimated adhesion coefficient and creep speed are calculated by fuzzy logic to get the adjustment coefficient of creep speed search; then the dynamic creep speed search method is used to get the reference creep speed needed for the current adhesion control, and finally, combined with the PI controller, the tracking of the train on the reference creep speed is realized through the control of motor torque. In order to verify the effectiveness of the designed method, a single-axle train traction dynamics model is established on the MATLAB/Simulink simulation platform for simulation verification. The results show that the search method can find the optimal creep speed quickly and accurately and realize the tracking control of high-speed trains under the switching of dry and wet track surface states.