Deformation prediction model for milling residual stresses in complex thin-walled parts with variable curvature

Complex thin-walled parts with variable curvature are crucial in aerospace, yet their manufacturing-induced residual stresses often lead to deformation, posing challenges in surface residual stress measurement and deformation prediction due to the complex geometric coordinate system. This study develops a model for predicting the distribution of subsurface residual stresses and the resulting deformation. By integrating the segmented polynomial fitting and random forest regression methods, a subsurface residual stress distribution prediction model is established and verified through experiments on machining and stress detection of such parts. Additionally, a special fixture for stress detection is designed. Based on ABAQUS software, a deformation simulation model considering residual stress is developed. Through continuous two-month monitoring of parts with different processing parameters, it is found that the simulation results align with the experimental trend, with an error range of 6.7 %–12.4 %. Eventually, a deformation prediction model based on process parameters is constructed, which can effectively predict the deformation of these parts and provides a theoretical foundation for deformation control.