The Finite Element Simulation Of Large-scale U-shaped Cross-section Beam Bending Process

The forming of large U-shaped cross-section beam with high-strength steel can causegreater spring-back, warping, lateral bending and other defects, and the high requirementsfor the precision of the size will increase the difficulty of the processing.Three different bending processes of large-scale U-shaped cross-section beam arepresented in this paper, with those processes the work-piece meeting the requirements canbe formed by single pass bending. The forming process is simulated with ABAQUS, themold structural and strokes of the punch and rolls applicable to the single pass bending ofhigh-strength steel are summarized by trial calculation and analysis to achieve a highforming accuracy. According to the results of the finite element simulation, the formingloads, the contour shape and the distribution of stress and strain of the sheet metal afterspring-back of those three different processes are analyzed, and the processes areoptimized and improved on this basis.As a result of the using of high-strength steel, there will be a greater spring-backwhile the forming of the sheet metal, which will greatly affect the development of themolds and the accuracy of the work-piece. The bending spring-back of the sheet metal is acomplex issue, and there are many factors coupled with each other which are highlynonlinear. Based on the characteristics of the U-shaped cross-section beam bending, a setof orthogonal experiments based on the finite element simulation are designed in thispaper. Through range analysis the factors affecting the spring-back radius are determined,they are the punch radius Rd, material parameter σs/E and the sheet thickness by theorder of their impacts on the size of spring-back radius. Based on this a mathematicalmodel of the spring-back radius of the sheet metal is deduced regression analysis, and theaccuracy of this model is verified. When the spring-back extent of the plate is very large,the accuracy of this model is poor. Study the spring-back radius with BP neural network, the prediction error is less than5%, indicating that BP neural network model can predictspring-back radius of U-shaped cross-section beam bending more accurately.