Modeling of Severe Shot Peening Effects to Obtain Nanocrystalline Surface on Cast Iron Using Artificial Neural Network

Abstract

Severe plastic deformation methods such as severe shot peening are used in order to improve mechanical properties of the components by surface microstructure nanocrystallization. Severe shot peening is one of the popular mechanical surface treatments generally aimed at generating nanograined layer and compressive residual stress close to the surface. Moreover, artificial neural network has been used as an efficient approach to predict and optimize the engineering problems. In present study effects of conventional and severe shot peening on cast iron were modelled by means of artificial neural networks and they were compared. The obtained results indicate that severe shot peening has superior effects on improvement of materials mechanical and metallurgical properties than conventional shot peening. Back propagation error algorithm and data of experimental tests on nodular cast ironwith ferrite-pearlite matrix were employed to train networks. Neural network testing was carried out using different experimental data which they were not used during networks training. In present paper distance from the surface is regarded as an input parameter and microhardness, residual stress and full width at half maximum are gathered as output parameters. The predicted values for full width at half maximum, microhardness and residual stress have the least statistical errors, respectively. The results demonstrate that there is a reasonable and applicable agreement among the experimental and predicted values, illustrating that the developed neural network can be employed to model the conventional and severe shot peening processes.