Abstract: Ti-15-3 alloy is a new metastable β- type titanium. The influence of hot deformation parameters on the microstructure of Ti-15-3 alloy after solution treatment has been studied by isothermal compression tests as well as quantitative metallographic analysis. On the basis of the data obtained from the tests, predicting models for equivalent grain size and recrystallization volume fraction have been established with an artificial neural network method. An FE numerical simulation system has been developed to simulate the distribution of microstructure in Ti-15-3 alloy after hot back extrusion and solution treatment by combining the neural network model with thermal-mechanical coupled rigid-viscoplastic FE model. Corresponding experimental research is performed. The agreement of the simulated results with measured ones shows that the simulation system is reliable.

Key words: Ti-15-3 alloy, Neural retwork, FE model, Simulation