Prediction of flow stress of Ti-15-3 alloy with artificial neural network

Hot compression experiments were conducted on Ti-15-3 alloy specimens using Gleeble-1500 Thermal Simulator. These tests were focused to obtain the flow stress data under various conditions of strain, strain rate and temperature. On the basis of these data, the predicting model for the nonlinear relation between flow stress and deformation strain, strain rate and temperature for Ti-15-3 alloy was developed with a back-propagation artificial neural network method. Results show that the neural network can reproduce the flow stress in the sampled data and predict the nonsampled data well. Thus the neural network method has been verified to be used to tackle hot deformation problems of Ti-15-3 alloy.