Structural damage identification of subseabed shield tunnels based on distributed fiber optic sensors and information fusion

Subseabed shield tunnels are constantly subjected to erosion induced by seepage and corrosive ions and electrochemical effects of diffuse currents from subway traction locomotives during service, resulting in damage to the outer wall of the segment. However, the outer wall cannot be directly observed, making damage identification a challenge. In this research, an approach based on distributed fiber optic sensors (DFOSs) and information fusion is proposed to solve this issue. To monitor strain, DFOSs were installed along the circumference of the segment inner wall. The strain curvature was adopted to solve the adverse influence of the spatial resolution (SR) averaging effect in DFOSs. Using the strain and strain curvature as information sources, the strain damage probability index (SDPI) and strain curvature damage probability index (SCDPI) were developed. Based on Dempster-Shafer (D-S) evidence theory, the SDPI and SCDPI were further utilized to build basic probability assignment (BPA) functions and perform information fusion for damage identification. To validate the feasibility of this approach, model experiments were conducted. The analysis of the test data indicated that information fusion can suppress the independent features of the two information sources while emphasizing their common features. The SR averaging effect, which can cause data distortion and diminish or even mask the damage anomaly feature in strain, and multiple extrema in the strain curvature that can interfere with damage localization are simultaneously eliminated. The damage identification accuracy is improved, and the damage localization error for the segment outer wall can reach 0.79 cm.