Application Of Robust Design Based On The Dual-Kriging Model In The Sheet Metal Forming

As one of the important processing method in the modem industry, sheet metal forming is used to produce a variety of sheet metal parts and has been widely applied in aerospace, defense, automotive and other fields. With the continuous development of the numerical simulation technology, the cycle of the mold production and product cost can be shortened by using the finite element to analyze the sheet metal forming. However, for the more complex forming parts, numerical simulation takes too much time. At the same time, the quality of the forming parts will be below the standard, affected by the fluctuations of the process conditions and the material properties in the production process. Therefore, the robust design should be applied for the metal forming process. It will make the forming quality of the sheet be not sensitive to the process conditions, material properties and so on.Kriging surrogate model is applied to optimize the sheet metal forming and it can reduce the time of the numerical simulation and enhance the optimization efficiency. The theta value of the random distribution of the Kriging model is optimized with the grey relational analysis and it can improve the accuracy of Kriging model without increasing the other sample points. Improved Kriging model is applied to predict the forming defects of the square box in Numisheet93. Compared to the unoptimized Kriging model, the accuracy of the model is greatly improved.To make the forming quality of the sheet metal be not sensitive to the fluctuation of the process condition, the material properties and so on, the robust design based on the Dual-Kriging model is put forward. The controllable factors and noise factors which affect the forming quality are chosen and their corresponding fluctuation range is determined. The function of the controllable factors and noise factors and quality of the forming parts is established using the improved Kriging model based on the numerical simulation. The second layer Kriging model of the forming quality and variance is built based on the first layer Kriging model and Monte Carlo simulation according to the parameters fluctuation. The robust model is optimized with the multi-objective particle swarm optimization algorithm based on the crowding distance. The result shows that, compared to the traditional deterministic optimization, the fluctuation range of the quality of the forming part is effectively reduced and the robustness is improved.