Prediction of thermo-mechanical behavior during hot upsetting using neural networks

A thermo-mechanical model is developed to predict metal behavior during hot working operations. At first, a neural network model is trained to calculate flow stress of deforming metal as a function of temperature, strain and strain rate and then by coupling the neural network model and a thermo-viscoplastic finite element model, temperature and velocity fields during hot open die forging process are predicted. To examine the model, hot nonisothermal upsetting on a low carbon steel is performed while force–displacement behavior and temperature history during hot working are recorded. A good agreement is observed between the predicted data and the measured results.