TY - JOUR
T1 - Classification of concrete corrosion states by GPR with machine learning
AU - Wong, Phoebe Tin wai
AU - Lai, Wallace Wai lok
AU - Poon, Chi sun
N1 - Funding Information:
The authors would like to thank the following bodies: The Civil Engineering and Development Department (CEDD) of the Hong Kong SAR Government for providing the field sites (HKM, NP), the professional advice on machine learning provided by Dr Christoph Völker of BAM, Germany, the Civil and Environmental Engineering Department of HKPolyU with the preparation of some of the specimens, and the English editing service provided by Dr Mick Atha. This study was financially supported by the following projects: General Research Fund of the Research Grant Council of Hong Kong, Project PolyU/15216619 ‘Waveguide Characterization of Shallow Subsurface Damages in Infrastructure Materials by Dispersive Behavior of Ground Penetrating Radar Wave’; Smart Transport Fund of HKSAR Government, Project PSRI/14/2109/RA ‘Smart Assessment of Bridge Deck Efficiency and Safety in Hong Kong’
Funding Information:
This study was financially supported by the following projects: General Research Fund of the Research Grant Council of Hong Kong , Project PolyU/15216619 ‘Waveguide Characterization of Shallow Subsurface Damages in Infrastructure Materials by Dispersive Behavior of Ground Penetrating Radar Wave’; Smart Transport Fund of HKSAR Government , Project PSRI/14/2109/RA ‘Smart Assessment of Bridge Deck Efficiency and Safety in Hong Kong’
Publisher Copyright:
© 2023 Elsevier Ltd
PY - 2023/10/26
Y1 - 2023/10/26
N2 - The evaluation of rebar corrosion in reinforced concrete by using ground penetrating radar (GPR) and machine learning (ML) is a complex process. In this paper, a multi-variate method is presented. It uses full-volume data obtained from the amplitude domain in a regular GPR x-y scanning exercise, and the shape of the rebar's reflection to categorise different corrosion phases. This method allows multi-dimensional analysis with quantifiable GPR attributes. GPR data were extracted from the field and laboratory and then labelled according to the ground truths and reference specimens. A classic ML algorithm, logistic regression, was applied. The cross-validation accuracy (sensitivity and specificity) of individual corrosion phases was high (>99%), and the false alarm rate was low (<1%). This work shows that GPR as an evaluation tool can assess unseen data like doing blind tests. Nonetheless, continuous expansion of the training database is suggested to increase its diversity in the future.
AB - The evaluation of rebar corrosion in reinforced concrete by using ground penetrating radar (GPR) and machine learning (ML) is a complex process. In this paper, a multi-variate method is presented. It uses full-volume data obtained from the amplitude domain in a regular GPR x-y scanning exercise, and the shape of the rebar's reflection to categorise different corrosion phases. This method allows multi-dimensional analysis with quantifiable GPR attributes. GPR data were extracted from the field and laboratory and then labelled according to the ground truths and reference specimens. A classic ML algorithm, logistic regression, was applied. The cross-validation accuracy (sensitivity and specificity) of individual corrosion phases was high (>99%), and the false alarm rate was low (<1%). This work shows that GPR as an evaluation tool can assess unseen data like doing blind tests. Nonetheless, continuous expansion of the training database is suggested to increase its diversity in the future.
KW - Concrete corrosion
KW - Ground penetrating radar
KW - Logistic regression
KW - Machine learning
UR - https://www.scopus.com/pages/publications/85169595859
U2 - 10.1016/j.conbuildmat.2023.132855
DO - 10.1016/j.conbuildmat.2023.132855
M3 - Journal article
AN - SCOPUS:85169595859
SN - 0950-0618
VL - 402
JO - Construction and Building Materials
JF - Construction and Building Materials
M1 - 132855
ER -