Abstract: There are some defects in sheet metal forming optimization, such as one-step method combined with optimization algorithm results in low precision, incremental method directly combined with optimization algorithm results in low efficiency, and constructing high-precision agent model through traditional method requires a lot of simulation results of incremental method. In order to take the advantage of both one-step method and incremental method, variable complexity method is adopted to establish a response surface model for difference compensation between one-step method and incremental method through less-frequency test sample point data. New test sample points are constructed through one-step method and response surface model for difference compensation. Moving least-square agent model is constructed on the basis of both new and original sample point data. The established moving least-square agent model has high precision and efficiency. Particle swarm optimization is used for optimization solution of the said agent model. The method is applied to metal forming optimization of automobile inner panel, and the result of optimization improves the sheet metal formability. The numeral example indicates that this method has high precision and strong engineering practicability.

Key words: Moving least square method, Optimization, Sheet metal forming, Variable complexity model