Restoring Force Identification And Modelling For Nonlinear Structures With Time Series Data

Due to material deterioration and creep, over loading, lacking of maintenance as well as natural and man-made hazards, engineering structures inevitably have different levels of damage accumulation. It has been realized from frequent engineering accidents happened recently around the world that it is crucial to execute the real-time on-line monitoring, to evaluate and to forecast the health condition and carrying capacity of infrastructures. The development of general identification and modeling approaches is one of the main challenging topics for the structural damage detection, performance evaluation, and control system design for disaster reduction and relief. Even exciting achievements have been made in inverse analysis for dynamic system, most of the current vibration-based identification via eigenvalues and eigenvectors and their derivations are suitable for linear elastic system. Generally speaking, the damage initiation and development procedure are typical nonlinear and engineering structures should be nonlinear system with different levels of nonlinearity. However, the theory for identification for nonlinear system is far less established than them for linear system.A data-based identification and modeling approach suitable for linear and nonlinear system is presented, and the feasibility of the method for identification and modeling of the structural nonlinear property has been investigated by theoretical analysis, numerical simulation and experiment study. The main contents of this thesis include:1. A data-based identification method was provided and the theoretical fundament was described. The proposed method can be employed to identify the mass, stiffness and damping coefficients of a linear structure and the equivalent linear system of a nonlinear system. And the implementation of the method was described when limited excitations were employed.2. Numerical simulation on the performance of the method was carried out. A magnetorheological (MR) damper, a typical nonlinear device, was introduced to a frame structure to simulate nonlinearity, and the dynamic responses of the nonlinear system equipped with MR damper were determined with numerical integration. The hysteretic nonlinear performance of the MR damper was identified quantitatively and compared with its numerical model. Results showed that the proposed method can identify the damper force of MR damper accurately. 3. A multistory frame structure equipped with a nonlinear member (MR damper) was designed and produced. Using the test measurement time series, different hystereric performance of the MR damper forces when different currents were inputted were compared with the test measurements. Results showed that the approach was feasible.4. A general nonlinear restoring force modeling methodology using time series was proposed. The performance of the proposed method was verified with simulated data and test measurements. Results showed that the restoring force forecasted by the nonparametric model agree well with the simulated force and test measurements under different excitation cases.