Analysis of workability behavior of Al–SiC P/M composites using backpropagation neural network model and statistical technique

Abstract

Abstract This paper presents an artificial neural network (ANN) model for predicting and analyzing the workability behavior during cold upsetting of sintered Al–SiC powder metallurgy (P/M) metal matrix composites (MMCs) under triaxial stress state condition which is the multifaceted technological concept, depending upon the ductility of the material and the process parameters. The input parameters of the ANN model are the preform density, the particle size, the percentage of reinforcement and the applied load. The output parameters of the model are the axial stress, the hoop stress, the axial strain, the hoop strain, the instantaneous strain hardening index, and the instantaneous strength coefficient. This model is a feed forward backpropagation neural network and is trained and tested with pairs of input/output data. A very good performance of the neural network, in terms of good agreement with the experimental data has been achieved. As a secondary objective, quantitative and statistical analyses were performed in order to evaluate the effect of the process parameters on the workability and the plastic deformation behavior of the composites.