Abstract: Dynamic load simulation is one of the core technologies of indoor bench testing, and it is also an important way to evaluate the comprehensive performance of the test bench. Aiming at a hybrid vehicle bench, the dynamic loading technology research was carried out based on the frequency measurement accuracy of the speed pulse signal that the model C8051 F120 could achieve. The dynamic model of the vehicle drive train and the bench system model was established in Matlab/Simulink, the load simulation algorithm and the PI adaptive Kalman filter algorithm controlled by the inverse model were designed, and the corresponding controllers were built to analyze the different sampling frequency and frequency measurement accuracy impact of load simulation. The simulation showed that the maximum load dynamic torque error of the motor was 10% when the low-speed section was at a sampling frequency of 100 Hz, and the frequency measurement accuracy was 0.001% without filtering. The high-speed section was filtered at a sampling frequency of 100 Hz, the frequency measurement accuracy was 0.001%, the system response time was 0.11 s, and the motor loading torque error was 15%. The loading system could run smoothly and could achieve the purpose of dynamic loading, which verified the correctness of the load simulation algorithm. After getting the correct simulation results, code generation could be easily achieved, which provided a reference for the hardware-in-the-loop test of the bench.