Research On Inversion Method of Soil Mechanical Parameters Based on Deep Neural Network

DOI

10.2991/978-94-6463-793-9_83How to use a DOI?

Keywords

PSO-GA-BP; soil mechanics; parameter inversion

Abstract

The accuracy of soil mechanical parameters has an important impact on the numerical simulation results of foundation pit engineering. Since the soil parameters often change during the construction process, the parameters need to be dynamically corrected in order to improve the accuracy of numerical simulation. In this paper, a soil parameter inversion method based on PSO-GA-BP neural network is proposed to simulate the construction deformation of the foundation pit support structure and invert key mechanical parameters such as soil elastic modulus (E) and cohesion (c) by combining the finite element simulation with the learning and optimization ability of deep neural network. The deep foundation pit project of a subway station in Wuhan is taken as an example, and the feasibility and accuracy of the method are verified by using on-site monitoring data. The results show that the PSO-GA-BP neural network integrates the global search capability of particle swarm optimization (PSO), the population diversity maintenance characteristic of genetic algorithm (GA), and the local refinement fitting advantage of back-propagation (BP) network, and the inverted soil parameters significantly improve the numerical simulation accuracy, and the maximal relative error is reduced to less than 5% from 14.15% before correction, which provides a reliable basis for engineering design and safety assessment. The maximum relative error is reduced from 14.15% before correction to less than 5%, which provides a reliable basis for engineering design and safety assessment.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Yongjian Liu
AU  - Zhenyao Yin
AU  - Xueli Xiao
AU  - Lan Luo
PY  - 2025
DA  - 2025/07/28
TI  - Research On Inversion Method of Soil Mechanical Parameters Based on Deep Neural Network
BT  - Proceedings of the 2025 8th International Conference on Traffic Transportation and Civil Architecture (ICTTCA 2025)
PB  - Atlantis Press
SP  - 978
EP  - 987
SN  - 2589-4943
UR  - https://doi.org/10.2991/978-94-6463-793-9_83
DO  - 10.2991/978-94-6463-793-9_83
ID  - Liu2025
ER  -