As the working environment of permanent magnet synchronous motor (PMSM) is becoming more and more complex, it is of far-reaching significance to find a speed regulation strategy that is not subject to system model mismatch caused by the environment and has adaptive capabilities for non-traditional mechanism modeling. With the development of microcontroller performance, the application of advanced control algorithms with a large amount of calculation in PMSM control becomes possible. Based on the application of the improved generalized predictive algorithm in PMSM vector control speed control, this paper designs the corresponding control model simulation and experiment, and compares the advantages and disadvantages of the improved algorithm. Aiming at the weaknesses of PI control's poor ability to resist load disturbances, this paper introduces an improved Generalized Predictive Control (GPC) algorithm with a smaller calculation amount to improve the problem of excessive calculation of the GPC algorithm, which is verified and compared in the motor simulation model. In the motor simulation model, the stepped generalized predictive control(SGPC), the control effects of SGPC, IGPC and PID algorithms were verified and compared in the simulation model of motor vector control speed regulation. The motor load change will also produce disturbance, which will make the system model mismatch. The motor load torque observer is introduced to compensate the disturbance of the controller, and the effect of the torque observer is compared under the condition of load change and constant load. By comparing the simulation results of the motor servo control with or without the observer, it can be seen that the observer improves the control precision of the system. This paper also automatically generates code based on the model-based design method supported by MATLAB, and builds the PMSM vector control system model, including the controller based on the improved GPC, speed loop, current loop, Clark-Park transformation and inverse transformation module, SVPWM module, etc. Using Texas Instruments TMS320F28335 chip and development kit, to complete the SGPC algorithm test verification, to prove its feasibility. And by introducing RTDX technology, the function of real-time data transmission in the running of the motor vector control program is realized.