Calculation method study on spring back of automobile ceiling

Abstract

Spring back is a mainly quality defect in the process of automobile ceiling stamping. Though researchers have made a lot of work on spring back by analytical methods or tests, superficial understanding or single method exists in the research of spring back prediction and control. Furthermore, Neural Network technology has applied in spring back prediction, but the method of Back Propagation Neural Network is usually used. And other Neural Network methods are rarely used. There is lacking horizontal comparison of different modeling precision. So combination method of test design, finite element analysis and Neural Network modeling is adopted in paper. And six parameters, such as plate thickness, die gap, friction coefficient, die fillet, stamping velocity and blank holder force, are adopted as research objects. Then average spring back values are calculated by FEM simulation. And modeling samples are obtained from orthogonal test design. At last, three spring back prediction models are built by Back Propagation Neural Network Radial Basis Function Neural Network. Generalized Regression Neural Network separately. Based on comparison of different modeling results, high precision method is selected as a mainly method of spring back prediction.