Development of a neural network to predict the final geometry of forged rings after cooling

peer reviewedThe paper deals with an application of Artificial Neural Networks (ANNs) to predict the geometry of hot forged pieces after cooling. Different ANNs have been considered and evaluated; then a network with two hidden layers has been set up. Training and testing data have been obtained through calibrated numerical simulations of the cooling phase carried out with a finite element (F.E.) code. The good agreement between predicted and numerical results confirms the possibility of using well-trained neural network to foresee final dimensions of pieces after hot forging operations