Parameter And Load Identification For Nonlinear Structures Based On Improved Unscented Kalman Filter

Engineering structures inevitably encounter various load excitations during their service period.When a structure is subjected to low-level load excitations,the structural behavior conforms to the linear elastic assumption.However,when it is subjected to strong load excitations,such as earthquake and strong wind,the structure performs strong nonlinear dynamic behavior to a certain extent,and its vibration response also shows nonstationary characteristics.Nonlinear structural parameter and load identification by structural vibration response is of great significance for calibrating the analysis model and evaluate the actual operation state of the structure.This thesis aims to establishing the method of synchronous identification of nonlinear structural parameters and loads under the condition of incomplete response information,and providing a novel algorithm for state evaluation of nonlinear structures under the condition of unknown input and incomplete output information.Based on the unscented Kalman filter theory,the nonlinear structural parameters are first identified under the condition that the external excitation of the structure is known.A synchronous identification method of nonlinear structural parameter and load based on improved unscented Kalman filter is then proposed.The main research work of this thesis is as follows.(1)The Kalman filter,extended Kalman filter and unscented Kalman filter algorithms are well introduced.Especially,the unscented Kalman filter algorithm is emphatically discussed.Aiming at the difficulty of selecting the initial value of the traditional unscented Kalman filter,the global iterative unscented Kalman filter method based on the global iterative technique is proposed.In order to verify the feasibility and effectiveness of the proposed method,a three-degree-of-freedom Bouc-Wen hysteretic nonlinear structure and a bilinear base isolation structure are simulated as numerical examples.The numerical results demonstrate that the global iterative technique can avoid the influence of improper selection of initial parameter,enhance the stability for structural nonlinear parameter identification,and improve the accuracy.(2)In order to simultaneously identify the nonlinear structural parameter and the excitation load,the traditional unscented Kalman filter method is improved,and a synchronous identification method for structural nonlinear parameter and external load is proposed.In the process of nonlinear parameter identification,the external load is preliminarily estimated by using the current predicted values of structural responses.In order to reduce the influence of measurement noise effect,the Kalman filter is embedded in the unscented Kalman filter to synchronously optimize the measurement noise covariance matrix.In order to verify the effectiveness of the algorithm,a single-degreeof-freedom nonlinear structure and a multi-degree-of-freedom nonlinear structure are simulated as numerical examples.The numerical simulation results demonstrate that the proposed method has good robustness and can effectively identify nonlinear structural parameter and external load.(3)Finally,a shaking table test of a five-story steel frame structure is conducted to verify the proposed method.In the experimental study,several dampers were installed in the steel frame to simulate the nonlinearity of the structure.The nonlinear parameter and external load of the five-story tested steel frame structure are identified synchronously by using the measured vibration response.The experimental results also indicate that the proposed method can effectively identify the nonlinear parameter and external load with high accuracy.