Digital modeling and uniformity control of entire physical fields during die forging forming process of complex components

The physical fields during the forging process of high-end components are characterized by complex geometric shapes, multi-field coupling, and significant non-uniformity. Precise real-time simulation of the entire physical field is a challenging bottleneck issue for the realization of intelligent forging. This work proposed a generic method for digital modeling of entire physical fields during the forging process of complex components that combines discrete extraction and overall prediction of field information. First, finite element simulations of the forging process were conducted under varying process parameters (including friction factors, billet temperatures, die temperatures, and forging velocities) to construct a modeling dataset. The dataset contains deformation physical quantities (including deformation temperature, strain rate, equivalent stress and equivalent strain) at 100 feature points on the forging at each forging stage. Then, using the forging process parameters and the coordinates of feature points as inputs, rapid prediction models for various deformation physical quantities were developed using the gradient boosting regression algorithm. The models' errors were all less than 10 %, achieving a rapid prediction of entire physical fields with a response time in seconds. Finally, a genetic algorithm was used to optimize the forging process parameters for more uniform temperature and strain rate fields, synchronously considering both the uniformity at different locations (the spatial uniformity) and the uniformity from the beginning to the end of the forging process (time-dependent uniformity). The forging with uniformly distributed grain size and hardness was obtained under the optimized process. This work could provide insights and research references for the digital modeling and intelligent control of the forging process.