Prediction of Folding Defect in Transitional Region during Local Loading Forming of Titanium Alloy Large-scale Rib-web Component

Abstract

A prediction model for the folding defect in transitional region during local loading forming of titanium alloy large-scale rib-web component was developed by the combination of finite element simulation, space-filling Maximin Latin hypercube designs method and back-propagation neural network. The FE model of local eigen model of transitional region during local loading forming of large-scale rib-web component was established based on DEFORM-2D and validated by physical experiment. The Maximin Latin hypercube designs method was applied to design uniform experimental schemes with geometrical conditions and reduction amounts as experimental variables. After simulations of the designed experiments, the back-propagation neural network was used to synthesize the data sets (results of folding defect). Finally, a prediction model was established for folding defect judgment in transitional region under various geometrical conditions and reduction amounts during local loading forming of large-scale rib-web components. The predicted results are quite consistent with the results obtained from FE simulation and experiment.