The hexagonal phase with close packed structure has a obvious anisotropic characteristic, and the primary α phase content in TA19 titanium alloy bar at room temperature is as high as 70%. Therefore, the α texture has a significant effect on the mechanical properties of TA19 titanium alloy bar, and effective texture prediction can greatly reduce the production cost and improve the production efficiency, and it can also help to determine the mechanism of texture formation. In this paper, the multi-scale coupling method of macroscopic finite element model and mesoscopic visco-plastic self consistent model was used to simulate the forging process of large TA19 titanium alloy bar under close to the actual process conditions, and the β→α phase transition process was considered. At first, the simulated β deformation textures of the center, R/2 and edge of the bar above the phase transition temperature were obtained. And then, according to the Burgers orientation relationship, the laws of variant selection of different positions of bar during cooling were obtained by analyzing the β texture characteristics of different positions, and the α transformation textures were obtained. Finally, combining with the α texture characteristics at the center, R/2 and edge of the bar after transformation, the orientation evolution of α grains with different initial orientation was analyzed when different slip systems started, and α deformation textures were obtained. By comparing the final simulation results with the actual forging results, it was found that they are in good agreement. It illustrated that the model had good reliability in predicting the deformation and transformation textures of titanium alloy bars during forging, which had great significance for controlling and adjusting the texture of titanium alloy bars during forging.