Artificial neural network approach for predicting the mechanical properties of Al7475/Flyash/SiC hybrid composite

• Published in: Hyperfine interactions 2024-07, Vol.245 (1)
• Main Authors: Kakkassery, Joseph J, Srinivasa, Rao N, Sethu, Ramalingam P, Jeyakrishnan, S, Vijayakumar, S, Pradeep, A
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Brinell hardness; Correlation coefficients; Fly ash; Hybrid composites; Mechanical properties; Neural networks; Parameters; Particulate composites; Predictions; Silicon carbide; Stirring; Ultimate tensile strength
• ISSN: 0304-3843; 1572-9540
• DOI: 10.1007/s10751-024-01993-z

In this study, AMCs is made by incorporating fly ash (FA) and silicon carbide (SiC) particles into Al 7475 through the stir casting method. The fabrication of samples involves changing parameters such as stirring speed (ST), stirring time (SI), and the weight% of FA and SiC (RE). The pouring temperature is maintained at a constant 700 °C throughout the casting process. This investigation specifically employs an Artificial Neural Network (ANN) model to forecast the ultimate tensile strength (UTS) and Brinell Hardness (BHN) of the resulting composite. Utilizing a wide-ranging dataset, a well-suited model is established to validate mechanical properties, enabling predictions for unidentified data. The ANN architecture (3–10–2) is depicted with high correlation coefficients 0.99703 for training, 1 for validation and testing), and an overall coefficient of 0.99513. The network is evident in the comparison between predicted and experimental results, affirming the viability of ANN in modelling effective stir casting parameters.