New Time-domain Methods For Simultaneous Identification Of Structural Parameter And Excitation Based On Incomplete Measurements

Structural system is complicate and the structural input or outputinformation is usually polluted and difficult to measure in practical engineering.In these cases the structural parameters and loads are difficult to identify, such asill-posedness, convergence problem due to large calculation, the coupling impactfrom uncertain structural parameters and loads, etc. New methods in time domainare proposed in this paper for simultaneous identification of structural loads andparameters based on limited input and output information. Main contents are asfollows:(1) A two-stage time-variant structural system identification method isdeveloped based on incomplete structural acceleration responses. In this method,the time variant structural parameters and loads are simultaneously identified byusing Extended Kalman Filter and average acceleration discrete algorithm. Theresults from both simulations and experimental tests validate the correction of theproposed method.(2) New methods are further developed to simultaneously identify thenonlinear structural parameters and loads with Unscented Kalman Filter (UKF)、EKF and orthogonal decomposition algorithm. The structural excitation isdecomposed by orthogonal approximation. Then the structural parameters andorthogonal parameters of excitation decomposition are simultaneously identifiedby EKF or UKF. Several simulations are implemented to validate the feasibilityof the new methods.(3) Two methods are developed to improve the stability of KF algorithm.Firstly, the physical meanings of structural parameters are guaranteed by thelognormal distribution assumption of the target state variables. Secondly, thecomputer floating point error is mitigated by revising the state covariance at eachtime step. Finally, numerical simulations from UKF are utilized to study theeffectiveness of the proposed method.