LAUSR

In this investigation, the mechanical and microstructural properties of aluminum composites reinforced by different reinforcing particles including SiC, TiC, ZrO2, and B4C were optimized using neural network and NSGA-II. In order to obtain the best microstructural and mechanical properties of aluminum composites, different friction stir processing parameters such as rotational and traverse speed and different reinforcing particles type were used in order to fabricate composites. Results show that friction stir processing significantly affect Si particles size as well as dispersion and fraction of reinforcing particles at the stir zone. Moreover, reinforcing particle types influence the mechanical properties of composites due to difference in hardness and thermal expansion of each reinforcement as well as bonding quality between each reinforcement and aluminum matrix. In order to model the correlation between the friction stir processing parameters and microstructural and mechanical properties of the composites, an artificial neural network model was developed. A modified NSGA-II by incorporating diversity preserving mechanism called the ɛ elimination algorithm was employed to obtain the Pareto-optimal set of friction stir processing parameters. Finally, an approach based on TOPSIS method was applied for determining the best compromised solution from the obtained Pareto-optimal set.