In order to reduce the vibration localization of multistage blade-disk system of aeroengine compressor, the integral model and substructure model of multistage blade-disk system are established by using substructure modal synthesis method, and the dynamic characteristics of two models are analyzed and the accuracy of the substructure model is verified by comparison. Based on the substructure model, two optimization algorithms of the random wear mass and the snake optimization are proposed. The results show that the vibration amplitudes of the two models are in good agreement in the first 200 modes, and the substructure model meets the requirement of analysis accuracy. Both optimization algorithms can effectively reduce the degree of the vibration localization of the mistuned blade-disk system. The random wear mass algorithm can quickly reduce the amplitude of the blade. The snake optimization algorithm can find the global optimal solution with enough iterations. The results show that the maximum vibration amplitude and the vibration localization factor after optimization are reduced by 5.72 % and 12.0 % respectively by using the random wear mass algorithm, and that of 9.13 % and 16.7 % respectively by using the snake optimization algorithm. The analysis method presented in this paper takes into account the analysis efficiency and analysis precision, it can be used for dynamic analysis and vibration reduction optimization of large power machinery such as aeroengine and gas turbine, and provides a basis for further improving the reliability and service life of aircraft.