GPR data processing methods based on machine learning algorithm to detect the backfill grouting of shield tunnel

Zeng, Li;Xie, Xiongyao;Zhou, Biao
(2022) 19th International Conference on Ground Penetrating Radar, Golden, Colorado, 12–17 June 2022 — Location: Golden, Colorado, USA (12.June.2022)

Files

gpr2022-1551.pdf
  • Open Access
  • Adobe PDF
  • 990.12 KB
Download

Details

Authors
  • Zeng, LiUCLouvain
    Author
  • Xie, Xiongyao
    Author
  • Zhou, Biao
    Author
Abstract
Shield tunnel method is currently the most important method for tunnel excavation in soft soil areas. With the construction and operation of a large number of tunnels in China, a certain degree of structural disease have appeared in the tunnels. How to control the settlements of tunnels and take corresponding measures to ensure the safety of operations has attracted great attention from design and operation management departments. In this chapter, FDTD numerical simulation of backfill grouting by GPR is performed. The GPR images of grouting layers with different thicknesses of 400MHz antenna and 900MHz antenna were simulated respectively, and the theory of integrated learning XGBoost is established. The XGBoost model was trained with numerical modeling data to obtain classification models of grouting layers of different thicknesses under 400 MHz and 900 MHz GPR, and the results showed that the GPR data could be classified and pattern recognized well under this circumstance
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Zeng, L., Xie, X., & Zhou, B. (2022). GPR data processing methods based on machine learning algorithm to detect the backfill grouting of shield tunnel. 19th International Conference on Ground Penetrating Radar, 13 Oct 2022, 166. https://doi.org/10.1190/gpr2022-155.1 (Original work published 2022)