A Comparative Analysis of ANN and taguchi for Enhancing Predictive modelling and optimisation for Al-Base Metal Matrix Composites reinforced with nanoparticles of SiC

Abstract

In this work, a detailed research of wear resistance and frictional behavior improvement in the metal matrix composite of aluminum-based Metal Matrix Composite was performed. Experimentally, Al 7072 alloy composites reinforced with SiC were taken for the fabrication process through stir casting method. The dry sliding wear test was performed and the factors L, S and C were varied from their minimum and maximum values and studied the effects on Sw of specific wear rate, and FF of friction force subsequently. Taguchi Design of Experiments Taguchi DoE provided a systematic way to explore the input parameter space and brought the optimal combinations as L=40N, S=30rpm, and C=9% to reduce minimum Sw and FF. In addition, Artificial Neural Network ANN model was created for the purpose of predicting the responses without doing experiments. A 10 hidden layer neuron ANN model results 100% accuracy through which the Sw and FF were calculated. Finally, Validation of optimal model result was also happened during with the experiments outcomes of the Taguchi model. The ANN model, linear regression plot, and other parameters showed good competency in terms of the degree of accuracy. Through this, the experimental research and model validation process provides good work which predicts the wear resistance and friction behavior for MMCs.

Keywords

• wear resistance
• friction behavior
• metal matrix composites
• taguchi method
• artificial neural network