Experimental Investigation and Prediction of Tribological Properties of Titanium Carbide and Multiwalled Carbon Nanotubes Reinforced Aluminium Composites using Artificial Neural Network

In this study, the tribological properties of TiC particle and MWCNTs reinforced aluminium (Al7475) hybrid composite synthesized by stir casting method were investigated by experimental and artificial neural network (ANN) model. Al7475 metal matrix composites was produced with different wt% of TiC and MWCNTs. The composite samples were tested at 0.42 ms^{- 1}, 0.84 ms^{- 1} and 1.68 ms^{- 1} under three different loads  (10N, 20N and 40N). The results indicated that Al7475+10%TiC+2%MWCNTs composite exhibit lower wear rate and reduced coefficient of friction in compare to other samples. TiC percent, MWCNTs percent, applied weight, sliding speed and Time were used as input values for the theoretical prediction model of the composite. coefficient of friction and Wear loss were the two outputs developed from proposed network. Back propagation neural network with 5 – 6 – 2 architecture that uses Levenberg –Marquardt training algorithm is used to predict the coefficient of friction and wear loss. After comparing experimental and ANNs predicted results it was noted that R² was 0.992 for wear loss and 0.980 for coefficient of friction. This indicated that developed predicted model has a high state of reliability.