View in EDS

Global maximum horizontal principal stress orientation: A high-precision machine learning framework based on multi-source heterogeneous data fusion.

Saved in:
Bibliographic Details
Title: Global maximum horizontal principal stress orientation: A high-precision machine learning framework based on multi-source heterogeneous data fusion.
Authors: Song, Zebin1,2 (AUTHOR), Jiang, Quan1 (AUTHOR) qjiang@whrsm.ac.cn, Zhang, Shishu3 (AUTHOR), Xia, Yong3 (AUTHOR), Li, Long1 (AUTHOR), Liu, Jian1 (AUTHOR)
Source: Engineering Geology. Sep2025, Vol. 356, pN.PAG-N.PAG. 1p.
Subjects: Machine learning, Multisensor data fusion, Earth sciences, Geotechnical engineering, Transformer models, Stress concentration, Structural geology
Abstract: Understanding the continuous spatial distribution of in-situ stress orientations is essential for safe and efficient underground engineering; however, traditional measurement methods are constrained by prohibitive costs and time requirements. We present a novel AI-driven framework that fuses neighborhood stress orientation patterns with topographic features through Vision Transformer (ViT) architecture and multi-layer attention mechanisms. This approach enables the first continuous, high-precision global mapping of maximum horizontal principal stress (S Hmax) orientations. Trained and validated on 32,464 quality-controlled records, the model extracts latent spatial stress orientation patterns across diverse tectonic settings, achieving 77.4 % prediction accuracy. Validation across four tectonically distinct regions confirms the framework's robustness, including its successful application to the Sichuan-Tibet Railway corridor where prediction errors reached as low as 1.3°. This approach overcomes the spatial continuity and cost limitations inherent in traditional stress orientations characterization, revealing significant application prospects from infrastructure planning to fault activity prediction. Together, these results demonstrate that integrating heterogeneous geoscientific data within an artificial intelligence framework enables high-precision prediction of stress orientations, offering novel insights into the evolution of global tectonic stress fields. • Integrates multi-source data using Vision Transformer and attention mechanisms. • Achieves high-precision mapping of global S Hmax orientations for the first time. • Validated in four tectonic settings and along the Sichuan–Tibet Railway corridor. [ABSTRACT FROM AUTHOR]
Copyright of Engineering Geology is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
FullText Text:
  Availability: 0
Header DbId: egs
DbLabel: Engineering Source
An: 187810817
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Global maximum horizontal principal stress orientation: A high-precision machine learning framework based on multi-source heterogeneous data fusion.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Song%2C+Zebin%22">Song, Zebin</searchLink><relatesTo>1,2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Jiang%2C+Quan%22">Jiang, Quan</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> qjiang@whrsm.ac.cn</i><br /><searchLink fieldCode="AR" term="%22Zhang%2C+Shishu%22">Zhang, Shishu</searchLink><relatesTo>3</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Xia%2C+Yong%22">Xia, Yong</searchLink><relatesTo>3</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Li%2C+Long%22">Li, Long</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Liu%2C+Jian%22">Liu, Jian</searchLink><relatesTo>1</relatesTo> (AUTHOR)
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22Engineering+Geology%22">Engineering Geology</searchLink>. Sep2025, Vol. 356, pN.PAG-N.PAG. 1p.
– Name: Subject
  Label: Subjects
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Machine+learning%22">Machine learning</searchLink><br /><searchLink fieldCode="DE" term="%22Multisensor+data+fusion%22">Multisensor data fusion</searchLink><br /><searchLink fieldCode="DE" term="%22Earth+sciences%22">Earth sciences</searchLink><br /><searchLink fieldCode="DE" term="%22Geotechnical+engineering%22">Geotechnical engineering</searchLink><br /><searchLink fieldCode="DE" term="%22Transformer+models%22">Transformer models</searchLink><br /><searchLink fieldCode="DE" term="%22Stress+concentration%22">Stress concentration</searchLink><br /><searchLink fieldCode="DE" term="%22Structural+geology%22">Structural geology</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Understanding the continuous spatial distribution of in-situ stress orientations is essential for safe and efficient underground engineering; however, traditional measurement methods are constrained by prohibitive costs and time requirements. We present a novel AI-driven framework that fuses neighborhood stress orientation patterns with topographic features through Vision Transformer (ViT) architecture and multi-layer attention mechanisms. This approach enables the first continuous, high-precision global mapping of maximum horizontal principal stress (S Hmax) orientations. Trained and validated on 32,464 quality-controlled records, the model extracts latent spatial stress orientation patterns across diverse tectonic settings, achieving 77.4 % prediction accuracy. Validation across four tectonically distinct regions confirms the framework's robustness, including its successful application to the Sichuan-Tibet Railway corridor where prediction errors reached as low as 1.3°. This approach overcomes the spatial continuity and cost limitations inherent in traditional stress orientations characterization, revealing significant application prospects from infrastructure planning to fault activity prediction. Together, these results demonstrate that integrating heterogeneous geoscientific data within an artificial intelligence framework enables high-precision prediction of stress orientations, offering novel insights into the evolution of global tectonic stress fields. • Integrates multi-source data using Vision Transformer and attention mechanisms. • Achieves high-precision mapping of global S Hmax orientations for the first time. • Validated in four tectonic settings and along the Sichuan–Tibet Railway corridor. [ABSTRACT FROM AUTHOR]
– Name: AbstractSuppliedCopyright
  Label:
  Group: Ab
  Data: <i>Copyright of Engineering Geology is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.)
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=187810817
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1016/j.enggeo.2025.108276
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 1
        StartPage: N.PAG
    Subjects:
      – SubjectFull: Machine learning
        Type: general
      – SubjectFull: Multisensor data fusion
        Type: general
      – SubjectFull: Earth sciences
        Type: general
      – SubjectFull: Geotechnical engineering
        Type: general
      – SubjectFull: Transformer models
        Type: general
      – SubjectFull: Stress concentration
        Type: general
      – SubjectFull: Structural geology
        Type: general
    Titles:
      – TitleFull: Global maximum horizontal principal stress orientation: A high-precision machine learning framework based on multi-source heterogeneous data fusion.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Song, Zebin
      – PersonEntity:
          Name:
            NameFull: Jiang, Quan
      – PersonEntity:
          Name:
            NameFull: Zhang, Shishu
      – PersonEntity:
          Name:
            NameFull: Xia, Yong
      – PersonEntity:
          Name:
            NameFull: Li, Long
      – PersonEntity:
          Name:
            NameFull: Liu, Jian
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 09
              Text: Sep2025
              Type: published
              Y: 2025
          Identifiers:
            – Type: issn-print
              Value: 00137952
          Numbering:
            – Type: volume
              Value: 356
          Titles:
            – TitleFull: Engineering Geology
              Type: main
ResultId 1