Modeling of Laser Shock Processing Technology Using an Artificial Neural Network to Determine the Mechanical Properties of the Ti–6Al–4V Titanium Alloy

Laser shock processing (LSP) is an innovative technology for surface modification applying the generated fields of compressing residual stresses within the near-surface domain of the investigated materials. Such stresses arise as a result of penetration of the shock wave (caused by high-energy nanosecond pulsed lasers) into the material; those waves significantly improve the mechanical properties and the fatigue characteristics of the metal materials and alloys. In the present work, to predict the residual stresses and the microhardness in the Ti–6Al–4V titanium alloy processed by the LSP technology, we engage a new method based on an artificial neural network. Here, we selected the following laser impact parameters: a laser pulse energy of 3, 5, and 7 J and a laser spot overlapping degree of 10, 30, and 50%. We applied the four-layer artificial neural network; as the input parameters, we took the laser pulse energy, the degree of overlapping, and the depth from the free surface, whereas the residual stress and the microhardness are considered as the output parameters. We show that the developed artificial neural network model with the 3 × 10 × 10 × 2 network configuration provides the best correlation with the experimental data in prediction of the residual stresses and the microhardness of the materials studied. For the optimal model, we obtained the mixed correlation coefficient, R2; the average absolute error, Δ; and the RMS error, ε: for the residual stresses 0.997, 7.226, and 9.956, and for the microhardness 0.987, 2.632, and 3.321, respectively. We might conclude that the artificial neural network is a reliable method for predicting the mechanical properties of the laser shock processed materials under a shortage of experimental data.