In this study, a general nonlinear restoring force (NRF) identification approach using structural dynamic response measurements and complete excitations is proposed at first. In this approach, the least-squares technique is employed to identify the parameters of an equivalent linear system of the nonlinear structure model based on the external excitations and the corresponding response measurements. The proposed approach is developed when the structure to be identified is incompletely excited. Both of the approaches are validated with a 4-story frame structure equipped with smart devices of magneto-rheological (MR) damper to simulate nonlinear performance. The identified NRF of the structure is compared with the test measurements. Results show that the proposed data-based approaches are capable of identifying the nonlinear restoring behavior of engineering structures and have the potential to be employed to evaluate the damage initiation and development procedure of engineering structures under dynamic loads.