A New Approach to Preform Design in Metal Forging Processes Based on the Convolution Neural Network

This study presents an innovative methodology for preform design in metal forging processes based on the convolution neural network (CNN) algorithm. The proposed approach extracts features of inputted product geometries and utilizes them to derive corresponding shapes by employing weight arrays (filters) determined during convolutional operations. filters are progressively updated training process, emulating learning steps a process engineer responsible processes. system is composed multiple three-dimensional (3D) CNN sub-models, which can automatically individual 3D candidates. It also implies that surfaces preforms easily acquired, important industry. was validated applying it two-dimensional (2D) axisymmetric shapes, one-quarter plane-symmetric two other industrial cases. In all considered cases, achieved substantial reductions load without defects, proving its reliability effectiveness application