Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
1562814 | Computational Materials Science | 2010 | 11 Pages |
The titanium alloy TA15(Ti–6Al–2Zr–1Mo–1V) has been experimentally investigated to gain insight into the modeling of hot deformation via a microstructural evolution law. To predict grain size in the primary α phase of TA15 titanium alloy, computer code written in FORTRAN has been developed to exploit an improved back-propagation (BP) neural network model. A training paradigm inherent in this type of model was programmed in accordance with data from hot simulation experiments and quantitative metallurgic techniques. Results show that the models developed are highly accurate and estimation errors for the BP neural network model after training are less than 5%. With additional interface subroutines, the code for this model has been embedded in the software DEFORM. A simulation module for microstructure evolution has been established and proved to be feasible. Isothermal extrusion of the TA15 alloy has been simulated and grain size predicted.