Natchanun Angsuseranee, Ganwarich Pluphrach, Bhadpiroon Watcharasresomroeng, and Arkhom Songkroh

DOI: 10.14456/sjst-psu.2018.87

pp. 534 - 539

Abstract

Advanced high strength steels (AHSS) have been used extensively in the automotive industry to reduce weight and fuel
consumption. However, increasing the strength of a material leads to the reduction in formability and a high degree of
springback. Moreover, sidewall curl has been detected from U-bending operations of AHSS which caused problems in the
assembly line. The aim of this research is to compare the efficiency of springback and sidewall curl prediction of AHSS grade
SPFC980Y in the U-bending process by the finite element method and artificial neural network approach. Input data for the
prediction consisted of punch radius (R_p), die radius (R_d), and blank holder force (F_b). The back propagation neural network
model was trained by the springback values from a U-bending die experiment with 27 conditions. Efficiency estimations of
springback and sidewall curl prediction were considered from the root mean square error (RMSE). The results showed that the
finite element method was more efficient than the artificial neural network approach. The RMSE values from the finite element
method for springback and sidewall curl were 0.104 and 0.092, respectively.