Abstract
The age-hardening curves of micro-hardness measurements obtained for sheets of Al–3wt%Mg alloy under different temperatures, applied loads and dwell times showed leveling and pronounced oscillations, indicating instability and reflecting a competition between the effect of dynamic recovery or sub-structure coarsening and the effect of solute drag and precipitation hardening. An artificial neural network (ANN) and the Rprop training algorithm were used to model the nonlinear relationship between the parameters of the aging process and the corresponding micro-hardness measurements. The predicted values of the ANN are in accordance with the experimental data. A basic repository on the domain knowledge of the age-hardening process verified the expected effect of micro-hardness decrease by increasing any of the applied parameters.