Research On Robust Optimization Of Sheet Metal Forming Based On Kriging And Grey Relational Analysis

Sheet metal forming is an important branch of material processing fields, and applied in automobile, aerospace etc. With the quickly development of computer science and finite element analysis theory, numerical simulation of sheet metal forming has been becoming one of the most important methods. This method can decrease manually operation, experiment cost and increase work efficiency. But this method need long time to run the finite element analysis (FEA) model for optimization. With industry development, many new materials have been developed. These materials are characterized by light weight, high strength and corrosion resistance. Fracture often occurs without any obvious necking phenomenon preceding the forming limit for the materials. Therefore, the FLD cannot predict the forming limit based on the tensile instability or bifurcation theories. Objects and constraint function of optimization can't been obtained based on FLD. In addition, the stability of part's quality may be affected by many noises, such as material parameters, processing parameters and die precision, thus leading to a requiring of decreasing the sensitivity of optimum functions to the uncertainties. Multi objects, such as no fracture, no wrinkle, minimum springback, are required in sheet metal forming optimization. Weighted method of multi-object optimization can't deal with the optimization of sheet metal forming well. In order to solve the problems mentioned above, the thesis accomplishes researches in the following topics and obtained the corresponding results.In order to decrease the time of optimization in sheet metal forming based on numerical simulation, Kriging metamodeling is proposed for sheet metal forming. Kriging metamodeling is researched based on design of experiment (DOE), and the nonlinear implicit relationship of input parameters and responses is established. Therefore the gradient computation can been obtained based on the relationship. To order to consider anisotropy in Kriging, the parameter of relational function is optimized. The procedure of reliable computation is written based on Kriging metamodeling. The cases in application of nonlinear and performance functions prove this can decrease time. The Kriging metamodeling is established in the flanging of sheet metal forming, and the process is optimized by the Kriging metalmodeling. The optimum parameters are obtained.Sheet metal forming has many criteria to evaluate the quality. The multi-object optimization is put forward based on grey relational analysis (GRA) for sheet metal forming firstly. There are different methods for different object functions, that is, higher is better, lower is better and desired value. The weights are computed making use of entropy. In order to obtain the optimum parameters, analysis of variance (ANOVA) for the grey relational grade is implemented. In order to improve the FEA computation, a computer aided program is developed. The GRA method is applied in welded beam and square cup forming. The research results showed grey relational analysis can be applied in multiple response optimization designs, and the optimum parameters are obtained.Because parts'quality is affected by the variation existing in the design parameters, a robust multi-object model is proposed to improve the parts'reliability and robustness. In the model, the GRA method is applied to solve multi-objects. Therefore, the GRA based on robust model is put forward to multi-objects. The initial solutions are obtained making use of genetic algorithm, and the optimum solution is obtained by the GRA. The result of robust optimization for welded bead proves the method is feasible.In the multi-object optimization of sheet metal forming, a GRA method based on Kriging robust model is put forward firstly. The key control and noise factors are screened by the GRA. The minimum bias latin hypercube design is used in Kriging metamodel. At the same time, key control and noise factors are batched in the Taguchi inner and outer table. The Kriging metamodels of multi-objects are established by the corresponding object data obtained in FEA of sheet metal forming. The GRA method is applied in Kriging robust model of sheet metal forming. The method is applied in mild steel sheet forming of NUMISHEET'93, that is the square cup forming. The optimization result proves the method is feasible.The grey relational multi-object method based Kriging robust model is applied in height strength sheet (HSS). Ductile fracture criteria(DFC) is applied to HSS forming for predicting fracture and optimization because FLD can't predict the forming limit of the material that hasn't large contraction before fracture. The tension tests of dual phase steel DP590 are done to obtain mechanical property. The deep drawing experiments are done to obtain processing parameters. The finite element model of rectangular cup drawing is verified by experiments. The constant of Cockroft DFC is obtained by the limit stroke in drawing experiment. The fracture in rectangular cup drawing is predicted by Cockroft DFC successfully. Object function of predicting fracture is established by cockroft DFC in complicated HSS forming. Fender forming processing is optimized based on the Kriging robust model. The optimization result of fender HSS forming proves the robust method can not only be used to solve multi-object optimization, but also the reliability and robustness of parts are improved.