A numerical method was innovatively presented to achieve uncertainty quantification of airfoil considering the randomness of distribution and magnitude of manufacturing error. Airfoil parameterization achieved by NURBS(nonuniform rational Bspline) method and random sample was generated by Monte Carlo method based on error detection and statistics. And then the Kriging surrogate model was trained to get the influence of machining error on compressor performance. Result shows the influence of profile error is related to the working condition and the loss may increase by 0853% at most. Moreover

the aerodynamic performance is more sensitive to suction surface than to pressure surface. The conclusion can provide more reasonable machining technology requirement and criteria of blade evaluation.