Experimental Study On Springback And Its FEM Simulation Contrast In Sheet Matal Forming Process

Springback is one of the common defects in sheet metal forming.There are strong demands from the manufacturing industry for accurate prediction and effective control of springback in order to improve quality of forming parts and reduce production cycle. Since 1970s, finite element method(FEM) and computer simulation techniques have been widely used in numerical simulation of sheet forming, but FEM has lacked a complete understanding of how to calculate the stress state precisely in forming parts so far, Springback has been proven surprisingly difficult to simulate accurately in the world, therefore experimental study on springback is still necessary. This paper emphases on experimental investigations on springback in typical forming operations such as convex edge flanging, concave edge flanging, "U" shape component forming,and "V" shape component forming.Firstly, this paper introduces some basic theories about FEM simulation, and discusses main factors affecting precision in springback calculation. Then experiments on springback in concave and convex edge flanging are carried out, significance of factors and their interactions are studied with statistics approach. An empirical formula for experiments is obtained using genetic algorithms, which can provide compensatory values for springback reduction. Furthermore, the mathematic model of springback is developed with radial basis function neural network based on the experimental data. Method of the mixed-level orthogonal test design is used to assist the selecting of training sets. Experimental investigations on springback in "U" and "V" shape bending operations are also carried out, and the corresponding artificial neural network is developed based on experimental data of "V" shape springback. Influences of main factors in "V" shape bending experiments on springback are illustrated by this neural network.Then compensation methods for these typical forming operations are analysed. At last typical forming operations in the experiments are simulated by ANSYS/LS-DYNA, and the simulation results are compared with experimental data.The paper investigates some main factors and their influence on springback, which can provide guidelines for reduction and compensation of springback in tool design and reference for FEM simulations.The use of artificial neural network in modeling of springback offer an effctive way to predict springback.This research isIIuseful for springback investigation and control in automotive covering parts.