Physical Parameter Identification Of Time-Varying Structure Using Partial Measurements

The identification of structural physical parameters plays an important role in the intuitive structural health assessment.In the field of structural health monitoring,vibration based techniques have been widely developed to identify the physical parameters of time-invariant structures.However,structures usually exhibit timevarying behavior due to severe hazards(e.g.strong seismic and wind loads)and other environmental effects(e.g.temperature or corrosion).It is of great engineering significance to describe the dynamic characteristics of time-varying structures and evaluate the performance of time-varying structural systems.There are still some difficulties and limitations in the identification of time-varying physical parameters,including:(1)the number of sensors is large,and even full observation of all the displacement,velocity,and acceleration responses is needed.Furthermore,the known external excitation is required;(2)A lack of effective identification methods for the large-scale time-varying structures.Although the substructure technique provides a workable path for the identification of large-scale structures,how to determine the interaction forces between adjacent substructures is still a great challenge in the case of inconvenient observation at the interface of substructures;(3)The identification of nonlinear time-varying physical parameters is more difficult and tedious than that of linear time-varying structures;(4)The identification of gradually varying physical parameters is more challenging than that of abruptly changing physical parameters;(5)The identification of the gradually-varying cable force is worth further study,etc.However,measuring the displacement,velocity and acceleration responses of all DOFs is unrealistic and uneconomic in practical engineering,and the external load information is often difficult to obtain directly.Therefore,it is suggested to investigate the identification methods for time-varying physical parameters by partially measured responses under unknown excitations,and explore suitable identification methods for large-scale structures,nonlinear systems,and gradually varying systems.It is more in line with the requirements of engineering practice,which has more theoretical research value and engineering significance.In this thesis,the followings are investigated step by step,including:"Physical parameters identification of linear time-varying structure under unknown excitations using partial measurements","Physical parameters identification of large-scale linear time-varying structure under unknown excitations using partial measurements","Physical parameters identification of nonlinear time-varying structure using partial measurements","Gradually-varying physical parameters identification of linear structure using partial measurements" and "The identification of gradually varying cable force using one acceleration response".Based on the Kalman filter series methods,the Kalman filter under unknown input series methods,and the substructure technique,the proposed methods improve the traditional methods based on the wavelet multiresolution analysis,and overcome the shortcomings of full observations of all displacement,velocity and acceleration responses.Additionally,the identification of time-varying physical parameters under unknown loads is discussed.In Chapter 1,the research progress of time-varying system identification at home and abroad is summarized,and the main research content and hierarchical structure are clarified.In Chapter 2,a novel method based on the Kalman filter under unknown input(KF-UI)algorithm and the wavelet multiresolution is proposed to identify the physical parameters of linear time-varying structures under unknown excitation.In Chapter 3,based on the substructure technique,a two-step method is proposed to identify the physical parameters of large-scale linear time-varying structures under unknown excitations by using partially measured responses.It overcomes the limitation that the proposed method in the second chapter is only suitable for small size timevarying linear structure.And observation at the interface of substructure is unnecessary.In Chapter 4,a two-step algorithm is proposed to identify the physical parameters of nonlinear time-varying structures by using partially measured responses.Based on the wavelet multiresolution analysis and unscented Kalman filter series methods,three cases are discussed respectively,including:the nonlinear structures with fewer elements under known loads,nonlinear structures with fewer elements under unknown loads,and nonlinear structures with more elements under unknown loads.In Chapter 5,a two-step algorithm is proposed to identify the gradually-varying physical parameters of linear structures by using partially measured responses.Combined with the discrete cosine transform(DCT)and Kalman filter series methods,the identification of gradually changed physical parameters under known and unknown excitations are discussed respectively.The proposed method overcomes the limitations of wavelet multiresolution analysis,such as generating a large number of scale coefficients when expanding gradually-varying parameters,dependence on scale function and scale level,and obvious boundary effect.In Chapter 6,a method based on discrete cosine transform and Kalman filter is proposed to identify the time-varying cable force by using one monitored acceleration response when the anemometer is installed on the bridge or not.In Chapter 7,the main work and innovations are summarized,and the future research work is prospected.