Abstract:The ABAQUS/ Explicit is used to analyze the wrinkling phenomenon of convex flange parts of 2A12O aluminum alloy in rubber forming. Numerical simulation tests of rubber forming were carried out with orthogonal test tools. By analyzing the effect of material performance parameters and process parameters on the forming parts of the convex flange, the main effect parameters affecting the wrinkling of the convex flange parts are obtained. Based on this experiment, a prediction model of wrinkling in BP neural network (BP NN) by the ant colony optimization (ACO) is established. After using many experimental data to train it, this model is applied by utilizing new data to predict the occurrence of wrinkling in convex flange parts, and this model was applied to predict the wrinkles of parts during the rubber forming. At the same time, the process test of 2A12O aluminum alloy sheet rubber forming convex flange parts is carried out. The results show that using this model can quickly obtain the optimal forming parameters when studying a new part forming, and it can be boosted the efficiency of its industrial. Last but not least, predictive errors are controlled within 5% and meets industrial application standards.