Forging-die Structure Intelligent Optimization Research Based On Surrogate Model

Forging is the commonly used processing method in metal plastic processing. Due to the material after forging have excellent mechanical properties, so this method widely applied in all kinds of mechanical products processing. As forging die is the main equipment in forging production, the quality of its design and manufacturing and service life determines the quality and cost of forgings. The quality of forgings mainly depends on the quality of forging die.The influence factors of forging process and forgings quality can be summarized as the mold structure and shape design, mould material, mould processing, forgings complexity, equipment performance, and other factors. The material flow during metal plastic deformation process mainly is influenced and controlled by the mold shape, therefore, reasonable selection and design of the mold structure shape parameters are particularly important.With the numerical simulation technology has become more mature, simulation-based design method in plastic forming technology has been widely used. This design approach is to calculate stress and strain in the metal forming process using finite element technology, and to determine whether the defects would have formed and to predict forming load, and then modify the process parameters and mold shapes through an intuitive analysis of metal forming process flow pattern and design variables on the impact of forming process in Post-processing results.In order to improve the forging die design efficiency and reduce manufacturing costs and improve product quality, it is necessary to optimize the forging process parameters in forging process and die structure that affect the quality of the forgings. At present, the optimum design method based on finite element analysis in forging process and its die design has become a trend.As one of the optimization method based on finite element analysis, the fitting optimization methods based on the objective function value is the method to be popularized currently because of its good general characteristics. The fitting optimization methods based on the objective function value, which is characterized by separation of optimization and the finite element program and versatility, can be directly use the existing commercial FEM software, give full play to its powerful finite element functions. The fitting optimization methods based on the objective function value is essentially agent model method. This method is to establish approximate model by fitting and to approximate the functional relationship between objective function and variables and then solving this approximate model to approximate the true extreme point.The key in the fitting optimization methods based on the objective function value is to establish the approximate model which can correctly reflect the relationship between the design variables and target function.In order to correctly reflect the importance of each parameter design variables, the reasonable experimental design method must be adopt to obtain the necessary sample points. After get enough sample point, the objective function value can be obtained using numerical simulation program for solving through a certain mechanism model. Then select the appropriate method to build approximate models. Finally, the approximate models were optimized analysis.Conventional optimization iterative method can not be used in the metal forming optimization problems because multi-factor,high-dimensional and non-linear. Intelligent optimization method is ideal for optimization of metal forming problems because it can not calculate derivative and the good global exploration ability. In addition, the Kriging model is suitable for high-dimensional nonlinear interpolation problems. This paper, the Kriging model coupled with genetic algorithms is proposed Kriging-GA optimization strategies for forging die structure parameters of optimal design. In this paper, Kriging-GA optimization strategies for optimal design of forging die structure parameters are proposed by the Kriging model and genetic algorithms coupled. Kriging-GA optimization strategy consists of three parts: building approximate model; the transformation of the multi-objective problem; genetic algorithm optimization. Kriging model Construction and genetic algorithm optimization program carried out by coupling using Matlab.Comparison with genetic algorithms, particle swarm algorithm has the advantages of easy to realize and fast convergence and few parameters need to be adjusted because it does not need genetic crossover and mutation operation. This article first propose Kriging-PSO optimization strategy coupled with kriging model and the particle swarm algorithms and achieve under Matlab programming.The Kriging-GA strategy and Polynomial response surface method were compared with applied in optimization for automotive flange forging die and crankshaft forging die. The results show that, Kriging-GA method has more accuracy but slow convergence than the polynomial response surface methods. On this basis, the Kriging-PSO optimization strategies applied to the crankshaft forging die as a contrast to the optimization problem. The results showed that results obtained with the Kriging-GA and with the Kriging-PSO had basically the same, but the Kriging-PSO convergence speed fast several dozen times. Finally, the Kriging-PSO optimization strategy used in cold extrusion of pre-perforated shells and end-forming combination of extrusion die of the optimal design to verify the Kriging-PSO optimization strategy is effective.