Artificial Neural Networks to correlate Hot Deformation Cooling Rate and Deformation Temperature on Continuous Cooling Transformation of 22MnB5 Steel

Abstract

The 22MnB5 steel is a hot stamping steel developed with the aim to satisfy the increasing request of the automotive industries to apply materials able to guarantee higher passive safety and weight reduction. The hot stamping process is an innovative forming technique in which the deformations are carried out at elevated temperature and allows to achieve high strength components. The experimental characterization of the material response, at different values of the main variables of process, may result both expensive and time consuming, but the mutual effects evaluation of the deformation parameters and the phase transformations are necessary to produce components within the desired properties. The developed model, by means of a neural network approach with a Bayesian framework, is able to predict the hardness and the specific microstructure of 22MnB5 steel as a function of the main parameters that are fundamental in hot stamping processes, thus overcoming the lack of fit of the existing numerical models.