Hyperspectral vision-based detection and random field modeling of age-dependent compressive strength of concrete segments

Abstract

The spatial distribution of compressive strength in shield tunnel concrete segments and its age-dependent dynamic evolution are vital to the safety and longevity of tunnel structures. However, traditional compressive strength assessment relies heavily on destructive testing, which is labor-intensive, time-consuming, costly, provides only localized information, and cannot capture the inherent spatial variability across the segment. To overcome these limitations and enable efficient, non-destructive, and spatially comprehensive evaluation, this study proposes a novel non-contact method integrating hyperspectral imaging (HSI) analysis with random field modeling (RFM). A hyperspectral camera (900–1700 nm) captured spectral data of concrete segments at various ages. Coupled with a proposed Spectral 3D ResNet deep learning model, this enabled high-precision compressive strength prediction, achieving an average cross-validation coefficient of determination of 0.918 and an average ratio of performance to deviation (RPD_cv) of 3.78, markedly superior to traditional models (partial least squares regression (PLSR), random forest regression (RFR), and convolutional neural network (CNN)). Further analysis revealed that segment strength development progresses through three distinct stages: the plastic stage (0–7 days) characterized by rapid growth but high variability and uneven distribution; the setting stage (7–15 days) with slowing growth and more concentrated distribution; and the hardening stage (15–33 days) where strength stabilizes with significantly enhanced uniformity and stability. Based on these findings, random field models of compressive strength at various ages were established, revealing the dynamic evolution of strength distribution from heterogeneity to homogeneity. At plastic ages, the random field exhibits pronounced spatial variability and longer correlation lengths; with increasing age, spatial correlation diminishes, resulting in a more uniform distribution with shorter correlation lengths at hardening ages. This study provides crucial theoretical backing and a data basis for non-destructive testing and comprehensive performance assessment of concrete shield segments, offering valuable guidance for quality control and structural design optimization in tunnel engineering.

Keywords

Shield tunnel / Concrete segment / Curing age / Compressive strength / Hyperspectral vision / Deep learning / Random field modeling

References

[1] Ahmed, M. T., Monjur, O., & Kamruzzaman, M. (2024). Deep learning-based hyperspectral image reconstruction for quality assessment of agro-product. Journal of Food Engineering, 382, 112223.

[2] Ali-Benyahia, K., Kenai, S., Ghrici, M., Sbartaï, Z.-M., & Elachachi, S.-M. (2023). Analysis of the accuracy of in-situ concrete characteristic compressive strength assessment in real structures using destructive and non-destructive testing methods. Construction and Building Materials, 366, 130161.

[3] ASTM (2013). C805/C805M-13a: Standard test method for rebound number of hardened concrete. ASTM International.

[4] Bas,yigit, C., Comak, B., Kılıncarslan, S., & Üncü, I. S. (2012). Assessment of concrete compressive strength by image processing technique. Construction and Building Materials, 37, 526-532.

[5] Botte, W., Vereecken, E., & Caspeele, R. (2023). Random field modelling of spatial variability in concrete - a review. Structure and Infrastructure Engineering, 1-14.

[6] Breiman, L . (2001). Random forests. Machine Learning, 45, 5-32.

[7] Caballol, D., Raposo, Á. P., & Gil-Carrillo, F. (2024). Non-destructive testing of concrete layer adhesion by means of vibration measurement. Construction and Building Materials, 411, 134548.

[8] Chamseddine, A. H., & Connes, A. (1997). The spectral action principle. Communications in Mathematical Physics, 186(3), 731-750.

[9] Chang, C.-I. (2003). Hyperspectral imaging: techniques for spectral detection and classification (Vol. 1). Springer Science & Business Media.

[10] Chen, Y., Jiang, H., Li, C., Jia, X., & Ghamisi, P. (2016). Deep feature extraction and classification of hyperspectral images based on convolutional neural networks. IEEE Transactions on Geoscience and Remote Sensing, 54(10), 6232-6251.

[11] Crespo, M. D., Molins, C., & Marí, A. R. (2013). Effect of variations in thermal-curing cycle on the cracking risk of precast segmental tunnel lining. Construction and Building Materials, 49, 201-213.

[12] Cusson, D., Lounis, Z., & Daigle, L. (2010). Benefits of internal curing on service life and life-cycle cost of high-performance concrete bridge decks-A case study. Cement and Concrete Composites, 32(5), 339-350.

[13] de B Harrington, P., Urbas, A., & Tandler, P. J. (2000). Two-dimensional correlation analysis. Chemometrics and Intelligent Laboratory Systems, 50(2), 149-174.

[14] Fan, L., Fan, M., Alhaj, A., Chen, G., & Ma, H. (2020). Hyperspectral imaging features for mortar classification and compressive strength assessment. Construction and Building Materials, 251, 118935.

[15] Fenton, G. A., & Griffiths, D. V. (2008). Risk Assessment in Geotechnical Engineering (Vol. 461). John Wiley & Sons New York.

[16] Ferrari, M., Mottola, L., & Quaresima, V . (2004). Principles, techniques, and limitations of near infrared spectroscopy. Canadian Journal of Applied Physiology, 29(4), 463-487.

[17] He, F., Liu, T., & Tao, D. (2020). Why resnet works? residuals generalize. IEEE Transactions on Neural Networks and Learning Systems, 31(12), 5349-5362.

[18] He, J., Hong, J., Gao, R., Tao, J., & Yan, H. (2024). Stochastic modelling of in-structure concrete strength based on stochastic damage model and Burr distribution. Structural Safety, 108, 102443.

[19] He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770-778).

[20] Huang, H., Wang, C., Zhou, M., & Qu, L. (2024). Compressive strength detection of tunnel lining using hyperspectral images and machine learning. Tunnelling and Underground Space Technology, 153, 105979.

[21] Huang, Y.-J., Zhang, H., Li, B.-B., Yang, Z.-J., Wu, J.-Y., & Withers, P. J. (2021). Generation of high-fidelity random fields from micro CT images and phase field-based mesoscale fracture modelling of concrete. Engineering Fracture Mechanics, 249, 107762.

[22] Kazemi, M., Madandoust, R., & de Brito, J . (2019). Compressive strength assessment of recycled aggregate concrete using Schmidt rebound hammer and core testing. Construction and Building Materials, 224, 630-638.

[23] Kingma, D. P. (2014). Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980.

[24] Kraskov, A., Stögbauer, H., & Grassberger, P . (2004). Estimating mutual information. Physical Review E-Statistical, Nonlinear, and Soft Matter Physics, 69(6), 066138.

[25] Lathuilière, S., Mesejo, P., Alameda-Pineda, X., & Horaud, R. (2019). A comprehensive analysis of deep regression. IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(9), 2065-2081.

[26] Lessard, M., Challal, O., & Aticin, P.-C. (1993). Testing high-strength concrete compressive strength. Materials Journal, 90(4), 303-307.

[27] Li, W., Wu, G., Zhang, F., & Du, Q. (2016). Hyperspectral image classification using deep pixel-pair features. IEEE Transactions on Geoscience and Remote Sensing, 55(2), 844-853.

[28] Li, Y., Yang, X., Tang, D., & Zhou, Z. (2024). RDTN: Residual Densely Transformer Network for hyperspectral image classification. Expert Systems with Applications, 250, 123939.

[29] Liang, H. (2012). Advances in multispectral and hyperspectral imaging for archaeology and art conservation. Applied Physics A, 106, 309-323.

[30] Liu, D., Cao, K., Tang, Y., Zhang, J., Meng, X., Ao, T., & Zhang, H. (2024). Study on weathering corrosion characteristics of red sandstone of ancient buildings under the perspective of non-destructive testing. Journal of Building Engineering, 85, 108520.

[31] Makantasis, K., Karantzalos, K., Doulamis, A., & Doulamis, N. (2015). Deep supervised learning for hyperspectral data classification through convolutional neural networks. IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 2015, 4959-4962.

[32] Ministry of Housing and Urban-Rural Development of the People’s Republic of China (MOHURD) (2019). GB/T 50081-2019: Standard for test methods of concrete physical and mechanical properties. National Development and Reform Commission of the PRC.

[33] Mulla, D. J. (2013). Twenty five years of remote sensing in precision agriculture: Key advances and remaining knowledge gaps. Biosystems Engineering, 114(4), 358-371.

[34] Olisah, C. C., Trewhella, B., Li, B., Smith, M. L., Winstone, B., Whitfield, E. C., Fernández, F. F., & Duncalfe, H. (2024). Convolutional neural network ensemble learning for hyperspectral imaging-based blackberry fruit ripeness detection in uncontrolled farm environment. Engineering Applications of Artificial Intelligence, 132, 107945.

[35] Rath, S., & Sakai, Y. (2023). Evaluation of the water penetration depth in mortar using water indicator and hyperspectral imaging. Construction and Building Materials, 380, 131269.

[36] Rebouças, J. P., Rohwedder, J. J. R., & Pasquini, C. (2018). Near infrared emission spectroscopy for rapid compositional analysis of Portland cements. Analytica chimica acta, 1024, 136-144.

[37] Reiter, L., Wangler, T., Roussel, N ., & Flatt, R. J. (2018). The role of early age structural build-up in digital fabrication with concrete. Cement and Concrete Research, 112, 86-95.

[38] Shen, D., Shi, X., Zhang, H., Duan, X., & Jiang, G. (2016). Experimental study of early-age bond behavior between high strength concrete and steel bars using a pull-out test. Construction and Building Materials, 113, 653-663.

[39] Uva, G., Porco, F., Fiore, A., & Mezzina, M. (2013). Proposal of a methodology for assessing the reliability of in situ concrete tests and improving the estimate of the compressive strength. Construction and Building Materials, 38, 72-83.

[40] Wahab, A., Aziz, M. M. A., Sam, A. R. M., You, K. Y., Bhatti, A. Q., & Kassim, K. A. (2019). Review on microwave nondestructive testing techniques and its applications in concrete technology. Construction and Building Materials, 209, 135-146.

[41] Wei, Q., Shen, L., Dunai, L., Kövesdi, B., Elqudah, S., & Cao, M. (2024). Quantitative evaluation on the effects of the spatial variability in concrete materials on seismic damage of concrete gravity dams. Engineering Fracture Mechanics, 307, 110287.

[42] Wise, J . (1955). The autocorrelation function and the spectral density function. Biometrika, 42(1/2), 151-159.

[43] Wold, S., Sjöström, M., & Eriksson, L. (2001). PLS-regression: A basic tool of chemometrics. Chemometrics and Intelligent Laboratory Systems, 58(2), 109-130.

[44] Xu, T., & Li, J. (2018). Assessing the spatial variability of the concrete by the rebound hammer test and compression test of drilled cores. Construction and Building Materials, 188, 820-832.

[45] Xu, Y., Hou, J., Zhu, X., Wang, C., Shi, H., Wang, J., Li, Y., & Ren, P. (2024). Hyperspectral image super-resolution with ConvLSTM skip-connections. IEEE Transactions on Geoscience and Remote Sensing, 62, 1-16.

[46] Yang, H., Huang, G., Chen, C., Yang, Y., Wang, Q., & Dai, X. (2024). Method for evaluation of geological strength index of carbonate cliff rocks: Coupled hyperspectral-digital borehole image technique. Journal of Rock Mechanics and Geotechnical Engineering, 16(10), 4204-4215.

[47] Zhang, J., Ma, G., Huang, Y., Aslani, F., & Nener, B. (2019). Modelling uniaxial compressive strength of lightweight self-compacting concrete using random forest regression. Construction and Building Materials, 210, 713-719.

[48] Zhuang, L., Ng, M. K., Gao, L., & Wang, Z. (2024). Eigen-CNN: Eigenimages plus eigennoise level maps guided network for hyperspectral image denoising. IEEE Transactions on Geoscience and Remote Sensing, 62, 1-18.