Experimental Investigation and Data Driven Prediction of Al6061-SiC-Gr Hybrid Composites

Al6061-based metal matrix composites are mostly used in structural and tribological properties due to their exceptional mechanical performance and lightweight nature. Reliable data-driven models are necessary since it is difficult to forecast composite characteristics accurately due to variations in processing circumstances as well as reinforcing content. The experimental examination and data-driven prediction of stir-cast Al6061-SiC-Gr hybrid composites are the main objectives of this work. To assess their mechanical, tribological, and physical characteristics, composites containing different weight percentages of graphite and silicon carbide were created. Standard protocols were followed in the experimental measurement of density, porosity, hardness, tensile strength, elongation, and wear properties. The impact of unreinforced, single-reinforced, and hybrid composite material performance was thoroughly investigated. To forecast properties, a Depth Adaptive Deep Neural Network (DADNN) was created and combined with an Improved Chicken Swarm Optimization Algorithm (iCSOA). Statistical performance indicators were used to train, validate, and test the proposed model using experimental data. Results indicated that the proposed model provides high prediction accuracy with 0.012 MAE, 0.015 RMSE, 0.0009 MSE, 0.998 R2, 0.143 MAPE, and 0.011 SDE and outperforms conventional approaches in estimating the properties of composite materials.