Artificial neural network-based prediction of complete forming limit curves for steel in sheet metal forming

Abstract

Forming Limit Curve (FLC) is crucial for predicting material formability and preventing defects in the sheet metal forming industry. Traditionally, FLCs are determined through Nakajima and Marciniak tests, which assess the material's response to various strain paths until the initiation of localized necking. However, these methods can be costly, time-consuming, and sensitive to factors like friction. Alternative approaches have been developed to address these challenges, including theoretical models and empirical methods based on tensile test data. This study investigates the use of Artificial Neural Networks (ANNs) to model FLCs, with the goal of improving prediction accuracy and efficiency. Input data for the ANN models were derived from tensile tests, incorporating parameters such as yield strength, ultimate tensile strength, uniform elongation, total elongation, normal anisotropy coefficient, and strain hardening exponent. The ANN models were trained to predict both FLC₀ and the complete FLC, and their outputs were compared with experimentally measured FLCs from Nakajima tests and empirical formulas from the literature. The results indicate that ANN techniques have significant potential to enhance the reliability and efficiency of FLC prediction.