Predictive Modeling of the Mechanical Properties of Alpha Alumina Using Artificial Neural Networks and Multiple Linear Regression

Abstract

In the present study, we built predictive models of the mechanical properties (Young’s modulus, fracture strength and toughness) of α-Al₂O₃. Experiments carried out on samples produced by spark plasma sintering (SPS). The experimental results were the basis for the evaluation of mathematical models and predictions by both the radial basis function neural network (RBFNN) and multiple linear regression (MLR) models. The results of the comparison of MLR and RBFNN models showed good agreement between the experimental data and the RBFNN model predictions whereas the MLR model reveals modest agreement with the studied mechanical properties.