Damage occurs in components and joints while the structure is affected by strong ground motions. Dynamic characteristics of the structure will change with the deterioration of strength and stiffness. Analyzing and processing the vibration signals is one of the mainstream ways for structural health monitoring and damage identification. In this paper, Hilbert-Huang transform is adopted to identify structural damage. Time-varying instantaneous frequency and instantaneous energy is used to identify the damage evolution of the structure. And relative amplitude of Hilbert marginal spectrum is used to identify the damage location of the structure. Finally, the acceleration records at gauge points from the shaking table test of a 12-storey reinforced concrete frame model are processed. Evolution and location of the model damage are identified. Identification results agree well with experimental observation. This indicates that the proposed approach is capable to identify damage of the structure.