ANN Based Prediction of Effect of Reinforcements on Abrasive Wear Loss and Hardness in a Hybrid MMC

Abstract

Problem statement: The reinforcements added to an alloy lead to variation in properties. The content and size of the reinforcement influences the properties of composites. Very little research has been carried out in hybrid composites. Work on hybrid LM6 aluminium alloy metal matrix composites (MMC) with flyash and SiC has been initiated here. The effect of the four parameters, size and weight of the reinforcements on the hardness and wear loss has been studied. Approach: Artificial neural networks, from the artificial intelligence family, is a type of information processing system, based on modeling the neural system of human brain. The effect of the parameters was investigated using ANN. Central composite rotatable method of design of experiments was used to arrive at the combination and the number of specimens. The specimens were prepared using the liquid metallurgy route and tested. Pin-on-disc apparatus was used for determining wear. Rockwell hardness on C scale was determined. The data from the experiments were used for training and testing the network. Results: The accuracy in ANN prediction was appreciable with the error estimated for wear loss and hardness being less than 2%. Conclusions/Recommendations: The ANN prediction is quick and economical way of estimating the properties.