<p>Thin-walled impeller blade curvature changes in the milling process. Low stiffness and other factors result in large milling processing errors. To compensate for online blade milling machining errors, this paper proposes a milling machining error prediction method by considering the curved surface features and deformation of the blade. First, based on the tool-workpiece contact relationship of blade curvature and machining deformation, undeformed and deformation chip thickness models are constructed to analyze the influence of curvature change and deformation on the chip thickness. Then, the variation of chip thickness in the cutter coordinate system is converted to the surface coordinate system, and the surface normal vector of the variation is taken as the predicted machining error. Finally, corresponding experiments are conducted on a five-axis machine, and experimental results indicate that the difference between the predicted machining error and the measured one is controlled within 21%.</p>