Artificial Neural Network Based Prediction Hardness of Al2024-Multiwall Carbon Nanotube Composite Prepared by Mechanical Alloying

In this study, artificial neural network was used to predict the microhardness of Al2024-multiwall carbon nanotube(MWCNT) composite prepared by mechanical alloying. Accordingly, the operational condition, i.e., the amount of reinforcement, ball to powder weight ratio, compaction pressure, milling time, time and temperature of sintering, as well as vial speed were selected as independent input and the mean micro-hardness of composites was selected as model output. To train the model, a Multilayer perceptron neural network structure and feed-forward back propagation algorithm has been employed. After testing many different ANN architectures, an optimal structure of the model i.e. 7-25-1 was obtained. The predicted results, with a correlation relation between 0.982 and 0.9952 and 3.26% mean absolute error, show a very good greement with the experimental values. Furthermore, the ANN model was subjected to a sensitivity analysis and the significant inputs affecting hardness of the samples were determined.

Keywords:

Al2024 Multiwall Carbon Nanotube , Composite , Artificial Neural Network , Microhardness , Mechanical Alloying