Prediction Flow Behaviour of a Magnesium Alloy in Hot Deformation and a Comparative Study by Using Hyperbolic Sine Model and ANN Model.

In order to get flow behavior of a deformation state magnesium alloy Mg–Gd–Y–Zr–Ag–Er, hot compression tests were performed on a Gleeble-3500 thermo-simulation machine ranging from 623 to 773 K and at strain rates of 0.001–1 s⁻¹. Hyperbolic sine model and feed-forward back propagation (BP) artificial neural network (ANN) model were used to predict flow behavior through different temperature, strain and strain rate. The correlation coefficient (R²) between the experimental and predicted flow stress in Hyperbolic sine model and BP-ANN model are 0.9700, 0.9996, respectively, the average relative error (AARE) corresponding to Hyperbolic sine model and BP-ANN model are 7.084 and 1.785%, the relative errors distribution range of BP-ANN model are more centralized than that of Hyperbolic sine model. Cross-validation approach was applied to analyze the predictability of the two models. The predictability of BP-ANN model is more accurate than Hyperbolic sine model. [ABSTRACT FROM AUTHOR]