Damage identification method on shield tunnel based on PLSR and equivalent damage analysis. (July 2023)

• Record Type: Journal Article
• Authors: Yan, Zhi-guo; Li, Jian-tao; Shen, Yi; Xiao, Zi-qi; Ai, Qing; Zhu, He-hua
• Is Part Of: Tunnelling and underground space technology. Volume 137(2023)
• Publication Date: 2023-07
• Subjects: Shield tunnel -- Damage identification -- PLSR -- Equivalent damage analysis -- Parametric study; Tunneling -- Periodicals; Underground construction -- Periodicals; Tunnels -- Periodicals; Underground areas -- Periodicals; 624.193
• DOI: 10.1016/j.tust.2023.105127
• Languages: English
• ISSNs: 0886-7798

Abstract

Graphical abstract: Highlights: Damage identification method based on PLSR is proposed. Proposed method has sufficient accuracy and sensitivity to identify damage. Input from circumferential strain has better sensitivity. Abstract: Damage identification plays a critical role in the maintenance of the shield tunnel structure. This study proposes a damage identification method for shield tunnel lining based on back analysis, which lays emphasis on the mechanical responses of the structure. The connection between mechanical parameters and the structural response is extracted by the finite element model. With the employment of Partial Least Squares Regression (PLSR), the load on shield tunnel lining can be determined by the monitoring data, the variation of which before and after structural damage, regarded as damage index, is used to identify the damage position. The equivalent damage load is defined to establish the equations of the load, deformation and damage, which solves the problems induced by the randomness of damage position. To clarify the validation, the method is verified by previous experimental results. Involving the structural damage of various positions and intensities, this method is applied with the numerical models based on circular shield tunnel. The parametric study shows the accuracy of the method under different damage conditions. The effects of monitoring data of radial displacement and circumferential strain were also compared. As a result, the primary input from the circumferential strain has better sensitivity than that of the radial displacement.