Investigation of damage identification of 16Mn steel based on artificial neural networks in tensile test

In order to identify the damage modes of 16 Mn steel, the tensile test of 16 Mn steel plate specimens was developed and monitored with acoustic emission technique. Based on the acoustic emission signature analysis during material damage, 3-layer back-propagation neural networks (BPNN) model with Tansig-Logsig transfer function was applied to identify the damage modes of 16 Mn steel in tensile test. In the model, amplitude, counts, energy, duration and rise time of acoustic emission parameters were selected as input neurons, and elastic deformation, yield deformation, strain hardening and necking deformation of damage modes were selected as output neurons. After the model has been trained with the experimental data, the discrimination rate of damage modes was equal to approximately 83%. It showed that it is feasible to identify the damage modes of 16 Mn steel in tensile test based on acoustic emission technique and artificial neural network.