Moving Force Identification Based On Nonnegative Flexible Conjugate Gradient Least Square Method And Experimental Study

Moving force identification(MFI)is an approach aims at acquiring a time history of moving vehicle loads,which has significance to bridge structure health monitoring and bridge designment.MFI deals with a typical second kind of inverse problem in structural dynamics.Since vehicle-bridge system has ill-possedness and noise cannot be avoided in responses,identification accuracy and identification efficient have been disturbed greatly,and the system cannot be solved by usual methods.In order to improve the identification accuracy,regularization methods should be used to obtain an approximate solution with acceptable error by imposing some constraints.This paper proposed a nonnegative flexible conjugate gradient least square(NN-FCGLS)method based on the mathematics modal of time domain method to identify moving vehicle loads.The proposed method is developed by enriching conjugate gradient least square method with flexible Krylov subspace technique and nonnegative constraint conditions.Therefore,the identification accuracy can be improved.Numerical simulation verified the performance of the NN-FCGLS method to identify the moving forces of the simple support beam,and experimental study are carried out to verify the effectiveness and applicability of the proposed method in MFI.Content and conclusions of the study as follows:(1)Numerical simulations of MFI for twin-axle vehicle loads under different noise levels and different response combination cases are carried out.The results show that the NNFCGLS method has advantages in robustness and ill-posed immunity under high noise level and all the response cases.And it is convenient to select iterative regularization parameters.(2)Numerical simulations of MFI for twin-axle vehicle loads with different speeds and different wheelbases are carried out.The results show that the NN-FCGLS method has good identification accuracy and robustness with different speeds and wheelbases.The proposed method is insensitive to speed and wheelbase,and can be used for MFI with different speeds and different wheelbases.(3)Numerical simulations of MFI for three-axle vehicles,five-axle vehicles and real car wheelbases are carried out.The NN-FCGLS method can effectively identify the moving forces of multi-axle vehicles under different noise levels and different response cases,which verifies the applicability of this method to identify moving forces of different vehicle types.(4)MFI experiments are performed on a simple supported beam bridge model.The identified forces of time domain method,conjugate gradient least squares method and the NN-FCGLS method are compared with the real axial loads,which shows that the NNFCGLS method can better identify moving forces.Responses are reconstructed by using identified forces of the NN-FCGLS method,and the reconstructed responses and measured responses are compared in both time domain and frequency domain.The results show that reconstructed responses can agree well with measured response in both time domain and frequency domain,which verifies the effectiveness and feasibility of the NN-FCGLS method for MFI.