Prediction of the mean grain size of MA-synthesized nanopowders by artificial neural networks

Abstract

In this work, mean grain size of synthesized nanomaterials (such as MoSi2–TiC and MoSi2–SiC), which were produced by mechanical alloying, has been modeled by artificial neural networks. A total number of 103 data were gathered from the previous works, trained, validated and tested by the built networks. The used data as inputs were the method of calculation of the mean grain size, milling time, annealing temperature, produced phases after mechanical alloying, vial speed and ball-to-powder ratio. The value of the output layer was the mean grain size. The obtained results from the testing phase of the trained networks showed that the models are capable of predicting the mean grain size of the mechanical-alloyed synthesized materials in the considered range.