A new data analysis technique, referred to the quadratic sum-squares error with unknown inputs (QSSE-UI) method along with adaptive tracking technique using genetic optimization algorithm, has been developed to simultaneously identify structural time-varying parameters and inputs based on real-time acceleration responses, leading to the detection and tracking of damages, including the time, location and severity of damages. The simulation results of 3-DOFs hysteretic nonlinear structure demonstrate that the proposed technique is capable of tracking the structural time-varying parameters and identifying the unknown inputs. Further, experimental study using a small-scale 3-story base-isolated structure model is conducted based on the proposed technique. During the tests, two typical excitations are applied to the base of the base-isolated structure model by shaking table, and different damage scenarios are simulated based on the stiffness element devices. Experimental results demonstrate that the proposed technique is capable of tracking the variation of structural parameters and simultaneous identifying the unknown support ex citations, leading to the accurate damage identification and tracking.