LiDAR-based structural health monitoring of drainage culverts in tailings ponds using an improved DBSCAN clustering method

Authors:
Publisher website
Google Scholar
Abstract
Structural health monitoring (SHM) of underground hydraulic structures such as drainage culverts in tailings ponds is critical for ensuring tailings-dam safety, yet conventional manual inspections and surface-based techniques struggle to capture internal deformations in dark, confined, and partially occluded environments. This study proposes a light detection and ranging (LiDAR)-based deformation-detection framework for SHM of drainage culverts that combines multi-temporal point clouds, cross-sectional geometric fitting, and an improved density-based spatial clustering of applications with noise (DBSCAN)-based clustering strategy. An initial LiDAR scan is used to construct a baseline three-dimensional model of the culvert, and subsequent LiDAR-simultaneous localization and mapping surveys are registered to this reference. Uniform cross-sections are extracted along the culvert axis, and Random Sample Consensus-based fitting provides reference functions to quantify radial deformation at the point level. Candidate deformation points exceeding a noise-informed threshold are then aggregated into physically meaningful deformation sections using a culvert-oriented adaptive DBSCAN strategy, which combines k -nearest-neighbor-based local neighborhood scaling, local-density-based core point screening, and engineering constraints on cluster size and axial span to improve robustness under uneven point density and partial occlusion. The framework is validated in a laboratory corridor and in the drainage culvert of the Luohe tailings pond. The method achieves mean absolute errors of approximately 6.68 and 9.67 mm, respectively, and reduces false deformation sections while correctly localizing all simulated deformation zones. These results demonstrate that the proposed approach provides robust localization and engineering-screening capability, with centimeter-level quantitative indication of deformation, for automated non-contact SHM of drainage culverts.
Funding Information
  • The Key Science and Technology Project of Ministry of Emergency Management of the People’s Republic of China (2024EMST141405)
  • the Natural Science Foundation of Shandong province (ZR2022QE251)
  • National Natural Science Foundation of China (51874268)
  • National Natural Science Foundation of China (52309132)

This publication has 32 references indexed in Scilit: