Prediction and compensation of stretch-bending springback for L-section profile

Abstract

This study aimed to achieve one-time stretch-bending forming in an L-shaped profile. Consequently, it focused on two main aspects: springback compensation and process parameter optimization. The process first combined simulation and experimental data, through a Co-Kriging model, facilitating the construction of an approximate model of a profile’s stretch-bending forming. A multi-objective optimization model was then formulated, to minimize springback displacement, deformation, and residual stress. The study employed the Non-dominated Sorting Genetic Algorithm II (NSGA-II) to explore the solution space for optimal combinations in stretch-bending processes. It created a single-objective optimization model to minimize the average positional error across each profile node. Subsequently, a simulated annealing algorithm was used to solve this model, enabling the determination of an optimal die profile, which met the required specifications of the one-time stretch-bending forming process. Moreover, a novel variable compensation factor method was proposed, which accurately determined the optimal compensation factors for each node, thereby enhancing node compensation precision.