Glass-Reinforced Aluminum Matrix Composite: Synthesizes, Analysis, and Hardness and Porosity Modeling Using Artificial Neural Networks

This study aims to investigate the predictive capabilities of artificial neural networks (ANNs) for Al-glass composites, specificallyin in exploring the effect of glass particle size and content on hardness, porosity, and microstructure in Al-glass composites. Contents between 0-15 wt.% glass particles, with two size ranges-less than 53 mu m, and between 53-75 mu m- were incorporated within the pure aluminum matrix. Powder metallurgy was employed to produce the composite specimens. Pressing at 400 MPa was applied to the powders to produce the green compacts. The sintering temperature was 550 degrees C. Three sintering periods were used: one, two, and four hours. The results indicate that the most significant factors affecting the hardness and porosity were glass percentage and sintering time. The highest hardness value of 27.50 HRB was obtained in specimen with 10% glass content sintered for 4 hours, with glass grain size of 53-75 mu m. Whereas the highest porosity percentage of 5.4% was recorded for specimen with 15% glass content sintered for 1 hour, with glass grain size of 53-75 mu m. For ANN, three inputs and one output were established, where the Levenberg-Marquardt training algorithm neural network had the highest accuracy of prediction. With highest value of R2= 99.96% and 99.99%, and RMSE=0.06855 and 0.007141 for hardness and porosity, respectively. As such a high prediction accuracy was obtained using the ANNs, this study proves that ANN is a significant tools for the prediction of nonlinear relationships. The novelty of this study lies in the combination of glass with aluminum as a new composite material, alongside the high predictive accuracy of the model with very small error margins, demonstrating the potential of ANNs to effectively handle nonlinear relationships in composite materials. Additionally, the ANN approach significantly saves time and costs associated with experimental testing and helps in finding the optimal combination with the best values of the mechanical properties, streamlining the development process for new composite materials. (c) 2024 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved