Damage Identification Of Bridge Space Structures Based On UKF

When bridges are affected by natural environment and human factors,structure damage may occur and accumulate over time.Before the serious engineering accident happened,it is necessary to identify structural damages and evaluate working state of bridges through structural monitoring devices.The accurate monitoring of the structure health will ensure the safety of the structure and prevent accidents.Among these damage identification methods,the unscented Kalman filter(UKF)based on UT-transform is a recursive model updating method.Compared with the traditional time domain method,UKF can obtain the optimal state estimation,compared with EKF and other methods,it do not need to solve the complex Jacobian matrix and have higher accuracy.Aiming at the ill-posedness of UKF in the inverse problem of structural damage identification,the pl regularization method is introduced into the UKF by the pseudo measurement technique.The proposed algorithm can effectively use different types of bridge damage prior information to improve the damage recognition results.And then,the regularized UKF is applied to the bridge damage identification with spatial characteristics.Specific work is as follows:(1)the state vector of UKF are constructed by using modal coordinates instead of structural nodes displacement,so the state vector dimension can be reduced by using modal truncation technique.Then,damage parameters are added to the state vector,and the mixed recognition problem of structure state and structural parameter are constructed.Using the free vibration response of the structure,the state estimation is obtained by recursive computation and the damage is identified.(2)The bridge damage information can be obtained by making full use of the manual inspection.Based on the pseudo measurement technology,an identification method of structural damage is presented by combining UKF with pl norm regularization.Different regularization methods can be chosen according to the damage characteristics of the structure,the ill-posedness of the inverse problem is effectively alleviated and the accuracy of damage identification is improved.In this paper,two regularization methods are compared through numerical examples.By analyzing the effect of different regularization methods and the damage condition,the effectiveness of the UKF method combined with 1l norm regularization is verified in the case of plane truss with local damage.(3)In the analysis of bridge structure damage identification,simply beam structure is the most common analysis model at present,but the actual bridge is three-dimensional structure with spatial characteristic.There are significant differences between the structural deformation,the damage zone and the layout of the measuring points.Two representative bridges,plate girder bridge and T-beam bridge,are chosen to study the damage identification of three-dimensional bridge structure.Firstly,the model of plate girder bridge with spatial characteristics is established by ANSYS,the effect of UKF combined with 1l norm regularization method on plate girder bridge is studied.What's more,it focuses on the influence of the spatial partition and the arrangement of the measuring point of the plate girder bridge on the damage identification.In the analysis of T-beam bridge,this article respectively regard the wet joints,transverse diaphragms,and the main beam as the damage objects and execute damage identification.At the same time,the effects of different modal information,numbers of partitions,and measuring methods on wet seam identification are also analyzed.For the spatial structure of the bridge,the numerical results show the UKF combined with regularization method can effectively identify damage,and has excellent robustness.