Stiffness identification of deteriorated PC bridges by a FEMU method based on the LM-assisted PSO-Kriging model. (September 2022)

• Record Type: Journal Article
• Authors: Wang, Xiaoming; Zhang, Jiading; Sun, Yuan; Wu, Zhoufan; Frederic C. Tchuente, N.; Yang, Fan
• Is Part Of: Structures. Volume 43(2022)
• Page Start: 374
• Page End: 387
• Publication Date: 2022-09
• Subjects: Structural health monitoring -- Static stiffness identification -- FEMU -- Static deflection -- PSO-Kriging; Structural engineering -- Periodicals; 624.1
• DOI: 10.1016/j.istruc.2022.06.060
• Languages: English
• ISSNs: 2352-0124

Abstract

Highlights: An efficient FEMU strategy is proposed for the stiffness identification of PC continuous beam bridges. The LM algorithm is firstly used to determine the sampling interval of the identification parameters. A PSO-Kriging model is used to promote the FEMU process. The effectiveness of the method is demonstrated through a full-scale test. Abstract: Stiffness identification from measured responses is an important part of structural health monitoring, in which finite element model updating (FEMU) is a widely studied method. However, the practicability of this method still needs to be improved when applied to large deteriorated civil structures. In this paper, a new FEMU strategy based on the PSO (particle swarm optimization)-Kriging model assisted by the Levenberg–Marquardt (LM) algorithm is proposed for the stiffness identification of deteriorated long-span prestressed concrete (PC) continuous girder bridges. The strategy is implemented in two phases: global stiffness identification based on the LM algorithm is adopted in phase I to determine the sampling interval of the identification parameters; then, in phase Ⅱ, the girder is divided into segments, and the accurate value of each segment stiffness is determined. The PSO-Kriging model is employed to facilitate the high accuracy and efficiency of the method. The effectiveness of the proposed method is demonstrated through comparison studies in an existing experiment and a full-scale test, and the parameter selection of the method is also discussed.