Research On Parameter Identification And Damage Detection For Nonlinear Hysteretic System Based On Adaptive Tracking Technique

The safe operation of civil,mechanical and other engineering structures directly related to the safety of people's life and property.However,in the long-term service process,these engineering structures will inevitably suffer from the occurrence and accumulation of structural damages due to some factors,such as the defects in the structures or function which may be presented in the early design and construction period,environmental erosion,material aging and load action by human or nature.These damages have a great threat to the safe operation of structures,and further,these damages may easily lead to catastrophic accidents.Early and timely structural health monitoring(SHM)can assist in the reprioritization of structural retrofitting schedules,extending the structural life,reducing testing and maintenance costs,and avoid catastrophic accidents.Besides,after severe natural disasters,such as earthquake,typhoon,etc.,the ability to quickly assess the safety and function of buildings and other civil engineering infrastructures is very significant for the efficient response of post-disaster,personnel rescue and emergency event management.Therefore,the advanced SHM technologies for the engineering structures are urgently required to online identify structural parameters and track structural damages based on the response data during the event.When the civil,mechanical and other engineering structures subjected to strong dynamic loads,these structures usually present high nonlinear characteristics,resulting in the hysteretic behavior,and the modeling for the nonlinear hysteretic structures and the system identification for these structures are very challenging in SHM area.High-damping rubber bearing isolation system is a typical project structure of nonlinear hysteretic system.In recent years,this isolation system has been widely applied on the building and other structures due to its advantages,such as strong damping effect,strong isolation effect,high seismic safety and low construction costs.However,currently,parameter identification and damage detection for nonlinear hysteretic system are still hot issues needed to be researched.As known,traditional system identification(SI)theory is based on the fact that all the system inputs and outputs information is known and then SI is used to identify the constant structural parameters.However,in practical applications,external inputs,such as seismic excitations,wind loads,etc.,may be not measured or even may be unmeasurable,but these unknown inputs,which reflect the interaction between the structure and surrounding medium,are very significant to the research of surrounding medium,such as foundation soil mass,wind,flow,etc..Besides,generally structural parameters typically varies with the damage occurrence and accumulation,such as the reduction of structural stiffness,and thus it is a very significant research to online detect and track structural damages when the damage occurred and accumulated during the SHM process.Consequently,to date,parameter identification and damage detection technology for the actual SHM application of engineering structures remains to be further developed though nowadays parameter identification and damage detection technology has made considerable progress.Based on adaptive damage tracking technique,three time-domain analysis techniques,which refer to as adaptive extended Kalman filter(AEKF),adaptive sequential nonlinear least-squares estimation(ASNLSE)and adaptive quadratic sum-squares error(AQSSE),have been investigated in this research to identify structural physical parameters and track structural damages.A series of numerical simulations and experimental verifications were conducted based on AEKF,ASNLSE and AQSSE method.Simultaneously,modeling for nonlinear hysteretic system and model simplification were discussed.Besides,an innovative device based on the concept of structural control,referred to as the stiffness element device,is proposed to online reduce the stiffness of structural components leading to structural local damage simulation.Based on the research above,three new data analysis techniques,which refer to as adaptive extended Kalman filter with unknown inputs(AEKF-UI),adaptive sequential nonlinear least-squares estimation with unknown inputs(ASNLSE-UI)and adaptive quadratic sum-squares error with unknown inputs(AQSSE-UI),have been developed to simultaneously identify structural time-varying physical parameters and unknown inputs based on adaptive damage tracking technique,leading to the detection of damages,including the time of instant damage occurrence,location and severity of damages.A series of numerical simulations and experimental verifications were conducted based on AEKF-UI,ASNLSE-UI and AQSSE-UI method.The main contents and achievements of this dissertation are as follows:(1)Parameter identification technique using extended Kalman filter(EKF)is investigated.Based on EKF method,an innovative adaptive damage tracking technique using constrained optimization is developed,and then,EKF using adaptive damage tracking technique is proposed,which refers to as adaptive extended Kalman filter(AEKF).A series of numerical simulations,including 3-DOF(Degree of Freedom)linear structure,3-DOF nonlinear structure and 3-DOF hysteretic structure,were conducted based on AEKF method.The simulation results show that AEKF method is capable of tracking the variation of structural parameters,leading to the detection of structural damages including the time of instant damage occurrence,location and severity of damages.Besides,based on the results of Bouc-Wen hysteretic model parameter sensitivity analysis and GZN110 natural rubber bearings testing,Bouc-Wen model can be further simplified.Then,experimental model of rubber-isolated structure was built,and using this experimental model,experimental verifications for parameter identification were conducted based on EKF method and three types of Bouc-Wen models with different number of hysteretic parameters.The experimental results show that model simplification is feasible and EKF method is effective and accurate.Finally,single base-isolated structure testing model was built for the experimental research.During the experiment,an innovative device,referred to as the stiffness element device,was applied to online simulate structural local damage.Then,AEKF method is used to track the variation of physical parameters of single base-isolated structure.The experimental results demonstrate that AEKF method is capable of tracking the variation of structural parameters leading to the detection of structural damages.(2)Parameter identification technique using sequential nonlinear least-squares estimation(SNLSE)is investigated.This method overcomes the drawbacks of EKF method,such as:(i)EKF method requires that the estimates of the initial values of the unknown parameters should not be far away from their theoretical values in order to obtain convergent solutions;(ii)Because of the large size of extended state vector,it involves considerable computational efforts,and SNLSE method has relatively high computational efficiency.Based on SNLSE method,an innovative adaptive damage tracking technique using constrained optimization is developed,and then,SNLSE using adaptive damage tracking technique is proposed,which refers to as adaptive sequential nonlinear least-squares estimation(ASNLSE).A series of numerical simulations,including 3-DOF linear structure,3-DOF nonlinear structure and 3-DOF hysteretic structure,were conducted based on ASNLSE method.The simulation results show that ASNLSE method is capable of tracking the variation of structural parameters,leading to the detection of structural damages including the time of instant damage occurrence,location and severity of damages.Besides,based on experimental model of rubber-isolated structure,experimental verifications for parameter identification were conducted using SNLSE method and three types of Bouc-Wen models with different number of hysteretic parameters.The experimental results show that model simplification is feasible and SNLSE method is effective and accurate.Finally,the experimental testing for single base-isolated structure testing model was conducted,and ASNLSE method is used to track the variation of physical parameters of single base-isolated structure.The experimental results demonstrate that ASNLSE method is capable of tracking the variation of structural parameters leading to the detection of structural damages.(3)Parameter identification technique using quadratic sum-squares error(QSSE)is investigated.This method overcomes the drawbacks of EKF method that mentioned above,and overcomes the drawbacks of SNLSE method,such as:Estimation of state vectors using Newmark-? method depends on the sampling frequency,etc.,and QSSE method has relatively fast convergent speed and relatively high computational efficiency.Based on QSSE method,an innovative adaptive damage tracking technique using constrained optimization is developed,and then,QSSE using adaptive damage tracking technique is proposed,which refers to as adaptive quadratic sum-squares error(AQSSE).A series of numerical simulations,including 3-DOF linear structure,3-DOF nonlinear structure and 3-DOF hysteretic structure,were conducted based on AQSSE method.The simulation results show that AQSSE method is capable of tracking the variation of structural parameters,leading to the detection of structural damages including the time of instant damage occurrence,location and severity of damages.Besides,based on experimental model of rubber-isolated structure,experimental verifications for parameter identification were conducted using QSSE method and three types of Bouc-Wen models with different number of hysteretic parameters.The experimental results show that model simplification is feasible and QSSE method is effective and accurate.Finally,the experimental testing for single base-isolated structure testing model was conducted,and AQSSE method is used to track the variation of physical parameters of single base-isolated structure.The experimental results demonstrate that AQSSE method is capable of tracking the variation of structural parameters leading to the detection of structural damages.(4)Based on AEKF method,the objective function and constraint equations of constrained nonlinear optimization method are built under unknown inputs,and then,the analytical solution for the adaptive extended Kalman filter with unknown inputs(AEKF-UI)is derived.A series of numerical simulations,including 3-DOF linear structure,3-DOF nonlinear structure and 3-DOF hysteretic structure,were conducted based on AEKF-UI method.The simulation results show that AEKF-UI method is capable of tracking the variation of structural parameters and simultaneous identifying the unknown inputs,leading to the detection of structural damages including the time of instant damage occurrence,location and severity of damages.Besides,3-DOF base-isolated structure testing model was built for the experimental research,and this model can simulate multiple damages in the form of different time of instant damage occurrence,different locations and different severity of damages during the experimental testing.Then,AEKF-UI method is used to track the variation of physical parameters of 3-DOF base-isolated structure and simultaneously identify the unknown support excitations in typical cases of vibration experimental testing,leading to the detection of structural damages including the time of instant damage occurrence,location and severity of damages.The experimental results demonstrate that AEKF-UI method is capable of tracking the variation of structural parameters leading to the detection of structural damages and identify the unknown inputs.(5)Based on ASNLSE method,the objective function and constraint equations of constrained nonlinear optimization method are built under unknown inputs,and then,the analytical solution for the adaptive sequential nonlinear least-squares estimation with unknown inputs(ASNLSE-UI)is derived.AEKF-UI method has the drawbacks similar with AEKF method mentioned above,and ASNLSE-UI method similar with ASNLSE method can overcome those drawbacks.A series of numerical simulations,including 3-DOF linear structure,3-DOF nonlinear structure and 3-DOF hysteretic structure,were conducted based on ASNLSE-UI method.The simulation results show that ASNLSE-UI method is capable of tracking the variation of structural parameters and simultaneous identifying the unknown inputs,leading to the detection of structural damages.Besides,the experimental testing for 3-DOF base-isolated structure testing model was conducted,and ASNLSE-UI method is used to track the variation of physical parameters of 3-DOF base-isolated structure and simultaneously identify the unknown support excitations in typical cases of vibration experimental testing,leading to the detection of structural damages,including the time of instant damage occurrence,location and severity of damages.The experimental results demonstrate that ASNLSE-UI method is capable of tracking the variation of structural parameters leading to the detection of structural damages and identify the unknown inputs.(6)Based on AQSSE method,the objective function and constraint equations of constrained nonlinear optimization method are built under unknown inputs,and then,the analytical solution for the adaptive quadratic sum-squares error with unknown inputs(AQSSE-UI)is derived.AEKF-UI and ASNLSE-UI method has the drawbacks similar with AEKF and ASNLSE method mentioned above,and AQSSE-UI method similar with AQSSE method can overcome those drawbacks of AEKF-UI method and AQSSE-UI method.A series of numerical simulations,including 3-DOF linear structure,3-DOF nonlinear structure and 3-DOF hysteretic structure,were conducted based on AQSSE-UI method.The simulation results show that AQSSE-UI method is capable of tracking the variation of structural parameters and simultaneous identifying the unknown inputs,leading to the detection of structural damages including the time of instant damage occurrence,location and severity of damages.Besides,the experimental testing for 3-DOF base-isolated structure testing model was conducted,and AQSSE-UI method is used to track the variation of physical parameters of 3-DOF base-isolated structure and simultaneously identify the unknown support excitations in typical cases of vibration experimental testing,leading to the detection of structural damages,including the time of instant damage occurrence,location and severity of damages.The experimental results demonstrate that AQSSE-UI method is capable of tracking the variation of structural parameters leading to the detection of structural damages and identify the unknown inputs.