Research on the Detection Method of Tunnel Surface Flatness Based on Point Cloud Data

Liufu Xiang, Yifan Ding, Zheng Wei, Hao Zhang, Zhenguo LiSymmetry 2021 · DOI

TL;DR

This paper introduces a novel method for accurately assessing tunnel surface flatness using 3D laser scanning (3DLS) technology. By developing a robust point cloud processing pipeline and defining both overall and local flatness metrics, the research offers a more efficient and precise alternative to traditional, manual inspection methods, crucial for ensuring tunnel construction quality and safety.

What Problem Does This Paper Address?

Ensuring the flatness of a tunnel's initial support is critical for the overall quality and long-term integrity of the tunnel structure. Uneven surfaces can lead to issues with waterproofing, subsequent lining stages, and potentially water seepage. Traditional methods for measuring this flatness, such as using a 2-meter ruler or total stations, are often slow, labor-intensive, and lack the required accuracy and efficiency for modern construction standards. This paper addresses the need for a more advanced, automated, and precise method for tunnel surface flatness detection.

Methods

The core of this research lies in leveraging 3D laser scanning (3DLS) to capture dense point cloud data of the tunnel surface. The proposed methodology involves several key steps:

  • Data Acquisition: Using a Topcon GTL-1000 3DLS to collect point cloud data from a 2-meter section of the tunnel's initial support.
  • Point Cloud Pre-processing: Employing SW Amberg Tunnel software for initial data cleaning and reduction, ensuring the preservation of surface characteristics.
  • Surface Reconstruction: Investigating and comparing four surface fitting methods (meshing, Poisson reconstruction, Lagrangian interpolation, and cubic B-spline interpolation) to find the most suitable for accurately representing the tunnel surface. The cubic B-spline interpolation method was found to be optimal due to its high accuracy, smoothness, and ability to retain local details.
  • Curvature-Based Smoothing: Applying a Savitzky-Golay (S-G) filter, incorporating curvature limitations based on design parameters of the Leicaoshan Tunnel, to further refine the fitted surface and ensure its authenticity and smoothness. This step helps to remove deviations from the actual surface while maintaining its shape.
  • Flatness Calculation Datum: Establishing a reference plane (flatness calculation datum) from the processed and smoothed point cloud data.
  • Normal Vector Distance: Calculating the distance from each original point cloud point to the flatness calculation datum plane along the normal vector.
  • Overall and Local Flatness Metrics:
    • Overall Flatness: Defined by the standard deviation of the normal vector distances, providing a measure of the overall unevenness of the tunnel surface.
  • Local Flatness: Introduced to quantify flatness at specific points or small regions. This method draws inspiration from the traditional 2-meter ruler method but is adapted for 3DLS data, considering the gradient direction and horizontal distance to define local deviations.
  • Flatness Distribution Map: Visualizing the calculated flatness across the tunnel surface to provide an intuitive understanding of uneven areas.

Datasets

The study utilized point cloud data collected from a 30-meter section of the Lushan Tunnel Project, specifically focusing on a 2-meter cross-section for detailed analysis. The data was collected using a Topcon GTL-1000 3DLS. No new benchmark datasets were introduced; rather, existing tunnel data was used for evaluation.

Key Insights and Ideas

  • 3DLS as a Superior Alternative: 3DLS technology offers significant advantages over traditional methods in terms of speed, accuracy, and non-contact operation for tunnel flatness detection.
  • Cubic B-Spline for Tunnel Surfaces: Cubic B-spline interpolation is identified as the most effective surface fitting method for tunnel point cloud data, balancing smoothness, detail preservation, and accuracy.
  • Curvature Limitation for Authenticity: Incorporating curvature limitations in the smoothing process is crucial for ensuring that the fitted surface accurately reflects the actual tunnel geometry and meets engineering requirements.
  • Dual Flatness Metrics: The introduction of both "overall flatness" and "local flatness" provides a comprehensive assessment, allowing for both a general understanding of surface quality and detailed analysis of specific problematic areas.
  • Step Distance (X) Impact on Local Flatness: The choice of step distance (X) in the local flatness calculation significantly influences the results, with larger X values leading to larger local flatness values, approaching the overall flatness. This highlights the importance of selecting an appropriate step distance for accurate local assessment.

Key Results

  • The proposed 3DLS-based method achieved a high degree of accuracy, meeting specification requirements.
  • The overall flatness of the tested tunnel section was found to be 19.9 mm, which is within the allowable deviation of 30 mm specified by the "GB/T 50299-2018 Construction Quality Acceptance Standard for Underground Railway Engineering."
  • Local flatness analysis showed that as the step distance (X) increased, the local flatness value also increased, stabilizing around 20 mm, which is comparable to the overall flatness.
  • Comparison with traditional total station methods indicated that the 3DLS method's true error met specification requirements (less than 25 mm), demonstrating its accuracy and effectiveness.

Limitations and Future Work

  • Datum Plane Optimization: The paper acknowledges that determining the "most suitable and optimal" flatness calculation datum plane requires further study.
  • Systematic Evaluation: The current research lacks a systematic evaluation of 3DLS-based flatness detection across various tunnel types and conditions.
  • Future Work: The authors suggest developing a more comprehensive 3DLS tunnel flatness detection and evaluation system as a future research direction. They also note that the current method is primarily designed for curved surfaces similar to tunnel interiors, and its applicability to flatter surfaces might require adaptation.

Summary updated

BibTeX
@article{xiang2021research,
  author = {Liufu Xiang and Yifan Ding and Zheng Wei and Hao Zhang and Zhenguo Li},
  title = {{Research on the Detection Method of Tunnel Surface Flatness Based on Point Cloud Data}},
  journal = {Symmetry},
  year = {2021},
  doi = {10.3390/sym13122239},
}

Loading related content...

Figure