The use of advanced metallic materials, including Advanced High-Strength Steels (AHSS), in the automotive industry has resulted in the development of advanced constitutive models to enhance the accuracy and reliability of finite element analysis (FEA) results. The Homogeneous Anisotropic Hardening model, based on distortional plasticity intend to accurately describe the plastic behavior of sheet metal materials, addressing phenomena such as the Bauschinger effect and permanent softening resulting from strain path changes. This study explores the modeling capabilities of Long-Short Term Memory structures (LSTM), a type of Recurrent Neural Networks (RNN) in predicting the material hardening parameters of a sheet metal material using the results of experimental three-point bending-unbending tests. The developed model takes the characteristic experimental punch force-displacement curve as input and provides the hardening parameters of the tested material as output. The data required for designing and training the network solutions were obtained from FEA, incorporating the HAH model, using a user material subroutine. The results indicate that the proposed LSTM-based approach performs comparably to traditional identification techniques in predicting material hardening parameters. This highlights the effectiveness of the developed procedure in characterizing hardening behavior under both monotonic and load reversal loadings for various materials, especially those extensively used in industrial applications, ranging from mild steels to AHSS.