A WGAN Approach to Synthetic TBM Data Generation

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review

Abstract

In this work we propose a generative adversarial network (GAN) based approach of generating synthetic geotechnical data for further applications in research and education. Geotechnical data generated by GANs shows similar characteristics as the original data, but still presents unique samples with no connection to the technical content of the original data. The data can therefore be made available publicly without any legal issues.

A WGAN (Wasserstein GAN) algorithm is used to generate synthetic tunnel boring machine (TBM) operational data based on real data from a major European tunnel construction site. The demands on the synthetic TBM data are of a dualistic nature: on the one hand, the data has to be sufficiently dissimilar to the original data, so that it does not create confidentiality issues (demand for originality). On the other hand, it has to show the same patterns and follow the same rules as the original data, so that it can be used as if it were real TBM data (demand for conformity). The WGAN model describes how a synthetic dataset is generated, in terms of a probabilistic model based on real data. By sampling from this model, we are able to generate new, unique synthetic and realistic TBM data.

We show that the demands for originality and conformity of the newly generated data are fulfilled.
Original languageEnglish
Title of host publicationTrends on Construction in the Digital Era - Proceedings of ISIC 2022
EditorsAntónio Gomes Correia, Miguel Azenha, Paulo J.S. Cruz, Paulo Novais, Paulo Pereira
PublisherSpringer, Cham
Pages3-19
Number of pages17
Volume306
ISBN (Electronic)978-3-031-20241-4
ISBN (Print)978-3-031-20240-7
DOIs
Publication statusPublished - 2023
EventInternational Society for Intelligent Construction 2022 Conference: Trends on Construction n the Post-Digital Era: ISIC 2022 - Avenida D. Afonso Henriques, 701, Guimarães, Portugal
Duration: 6 Sept 20229 Sept 2022
Conference number: 3
https://icisic2022.com/

Publication series

NameLecture Notes in Civil Engineering
Volume306 LNCE
ISSN (Print)2366-2557
ISSN (Electronic)2366-2565

Conference

ConferenceInternational Society for Intelligent Construction 2022 Conference: Trends on Construction n the Post-Digital Era
Abbreviated titleISIC 2022
Country/TerritoryPortugal
CityGuimarães
Period6/09/229/09/22
Internet address

Keywords

  • Synthetic data generation
  • Generative adversarial networks
  • Machine learning
  • TBM operational data
  • Tunnelling
  • Tunneling

ASJC Scopus subject areas

  • Civil and Structural Engineering

Fingerprint

Dive into the research topics of 'A WGAN Approach to Synthetic TBM Data Generation'. Together they form a unique fingerprint.

Cite this