Improvement And Application Of Extended Kalman Filter With Unknown Inputs

In recent years,more and more structural health monitoring systems have been installed in modern buildings with the rapid development of structural health monitoring technology,so that we can know the state of the structures and take timely measures to ensure the safety of the structures.The civil engineering structure is becoming larger and more complex,which puts higher requirements on structural health monitoring.How to accurately estimate the state and parameters of large structures with limited sensors has become a very challenging problem.Extended Kalman Filter(EKF)is an effective method for real-time identification of structural state and structural parameters using partially measured structural dynamic response.It has been widely used in structural health monitoring and vibration control research.However,the traditional EKF method is only applicable when the external inputs to the structure is known,and it is difficult to measure all inputs to a structure in the actual situation,which greatly limits the application of the extended Kalman filter method.In recent years,some scholars have proposed some improved extended Kalman filter methods(EKF-UI)with unknown inputs for simultaneous identification of structural parameters,structural states and unknown inputs.However,the proposed extended Kalman filter methods under unknown excitation still have some limitations,such as measurement position requirements for acceleration response,drifts in the estimated structural displacements and so on,which limits the application of these methods in practice.In addition,the acceleration measured by the acceleration sensor in actual situation is absolute acceleration,and the extended Kalman filter method with unknown inputs usually needs to measure the relative acceleration,which requires the ground acceleration information to be known.Therefore,the existing EKF-UI method is difficult to apply to situations where the ground inputs is unknown.This paper mainly carries out the following research work based on the above research background:The second chapter of the thesis proposes a data fusion based general extended Kalman filter with unknown inputs method(GEKF-UI).The method is derived on the basis of the traditional extended Kalman filter framework.It can identify structural state,parameters and unknown excitation in real time,and this method only needs to measure partial structural acceleration response and does not need to measure the acceleration at the location of external excitation.In addition,the method also effectively solves the low-frequency drift phenomenon in structural displacements and unknown excitation identify process through the fusion of acceleration and strain or displacement data.The third chapter of the thesis proposes the general extended Kalman filter with unknown ground input.In this method,only the absolute acceleration of the structure is measured,the structural state,structural parameters,unknown ground excitation can be estimated in real time.Data fusion of the absolute acceleration and interlayer displacement are used in this method to prevent the previous drifts in the estimated structural displacements and unknown inputs.The fourth chapter of the thesis combines the GEKF-UI method with the substructure identification method for the structural state and damage identification of large size structures.The boundary force of the substructure is treated as additional unknown inputs.Real time identification of structural states,parameters,unknown excitations can be achieved without substructure boundary force or substructure boundary response measured.This method can be well applied to damage identification of key parts of large size structures.The fifth chapter of the thesis applies the GEKF-UI method to structural multi-level damage identification.The damage identification is performed at two levels:the first is at large-scale element level to locate the potentially damaged region and then the large-scale damage element is accurately modeled to further identify the damage location and severity.The last chapter of the thesis reviews and summarizes the main work of this paper and the innovation of the thesis,and looks forward to the next research content.