Comprehensive prediction method for die-roll height of fine-blanking components

This study proposes an economic and effective approach for predicting the die-roll height of fine-blanking components by fully considering the effect of three types of parameters: geometric features of components, process parameters of fine-blanking, and material parameters of fine-blanked sheets. Two prediction submodels that involved geometric features and process parameters individually were established based on a back-propagation artificial neural network (BP ANN), merged into one model, and finally generalised to different materials by correction according to experimental results. During modelling, simulations rather than experiments were used to obtain data for supporting BP ANN considering the demand for a single controlling variable and for saving economic and time costs. Analyses showed that geometric features had a more distinct influence on die-roll height; furthermore, the effect of process parameters is complicated and can be described more suitably by BP ANN. The prediction results of the combined model were favourable compared with the experimental results. Furthermore, the experimental curve used for correction during generalisation of the model showed that a material with higher yield strength generally had smaller die-roll.