Artificial neural network prediction on ultrasonic performance of bismuth-tellurite glass compositions

Abstract Artificial neural networks (ANN) is known as one of the artificial intelligence tools which are inspired by the biological nerve system, have a capability to predict the physical and elastic parameter of glasses without melting the raw materials. The experimental of bismuth-tellurite glasses with the composition yBi2O3 - (1-y)TeO2 where y = 0, 0.05, 0.07, 0.10, 0.13, 0.15 have been fabricated using melting and quenching methods. These works were discovered that the prediction value by artificial neural networks for density, ultrasonic velocity, and elastic moduli of bismuth-tellurite glass composition gives a very good agreement as compared with the experimental measurements. The goodness of fit from the graph used R2 value to represent the relationship between the data presented from the experiment and prediction model. The great fit of coefficient R2 value elucidates in all figures is around 0.99942-1.0000 which is considered to be very satisfactory.