Volume : 3, Issue : 2, February - 2014

On the Use of Function Approximation Potential of Artificial Neural Networks for Predicting Elastic Moduli of Binary Oxide Glass Systems

Dr. R. Sheelarani, Dr. K. T. Arulmozhi

Abstract :

Function approximation (non-linear regression) is one important application of artificial neural networks (ANNs). Since the elastic moduli of multicomponent materials are dependent on the properties of the constituents in a nonlinear way, the ANN can be used to predict the elastic moduli of these systems. In this work a feed forward multiplayer perceptron (FFMLP) has been described to predict elastic moduli of binary oxide glasses using some of the characteristics of the component oxides as predictor variables.