This study characterizes the plastic behavior under cyclic loading for an aluminum alloy of 5182-O and advanced high strength steel of QP1180. Twin-bridge shear specimens were loaded cyclically to evaluate the reverse shear loading behavior of the tested materials with different pre-strain from small (0.05 major strain) to large (0.15). The plastic behavior under cyclic loading is modelled by kinematic and distortional hardening models under the non-associated flow rule (NAFR) by anisotropic Drucker (Aniso-Drucker) yield function. An additional shear constraint proposed by (Abedini et al., 2018) is imposed on the plastic potential of NAFR. The Zang and HAH20 models are selected as kinematic and distortional hardening models respectively. They are calibrated inversely by the reverse torque-torsion angle curves from experiments. Finally, V-bending tests with the pre-strained strips were carried out and simulated by the calibrated hardening model to evaluate its predicting accuracy. The results show that twin-bridge shear test combined with the recommended inverse engineering approach is implementable to overcome the drawback of the simple shear test due to inhomogeneous deformation in large strain. The constitutive models are critical for the inverse calibration. It is found that the calibration accuracy under NAFR is higher than the associated flow rule (AFR) especially for large strain conditions. Also, the additional shear constraint imposed on the plastic potential of NAFR is found to be an effective approach to enforce the simulated shear stress at zero hydrostatic pressure. As for hardening model, the HAH model has more flexibility compared to the kinematic hardening model benefiting from its distortional hardening form. Also, parameters related to permanent softening in the new HAH20 model need to be evaluated carefully at the large deformations.