AI-based damage detection in prestressed concrete beams: a vision-integrated deep learning framework for crack localization and severity classification

Abstract

Aging prestressed concrete (PC) structures, particularly those utilizing unbonded tendons, are susceptible to long-term deterioration mechanisms such as creep, shrinkage, and prestress loss, which manifest in complex damage forms including surface cracking and FRP debonding. Traditional inspection techniques are often labor-intensive and insufficiently accurate in detecting early-stage or subvisible defects. This study proposes a novel artificial intelligence (AI)-driven framework for automated defect detection and classification in prestressed concrete beams based on visual and structural response data. This work examines progressive damage in newly cast laboratory prestressed concrete beams under controlled loading, using synchronized high-resolution vision and vibration measurements, and evaluates specimen-disjoint generalization and edge-oriented deployability. A hybrid deep learning architecture combining YOLOv8 for crack localization, Swin Transformer for damage severity classification, and CNN–Transformer models for time-series vibration analysis is developed. The results demonstrate that the proposed system achieves a crack detection accuracy of 96.3%, a severity classification F1-score of 93.1%, and can localize damage with a mean IoU of 0.82. This work presents an integrated, aging-aware multimodal AI framework for damage assessment in prestressed concrete beams with unbonded tendons, combining real-time vision-based crack/debonding localization with synchronized vibration/strain analysis to provide a scalable alternative to traditional SHM methods.

Keywords

Prestressed concrete beams / Structural health monitoring (SHM) / Crack detection / Damage severity classification / YOLOv8 / Swin Transformer / CNN–Transformer / Multimodal deep learning / Vibration signal analysis / Aging infrastructure

References

[1] Abbasi M, Consilvio A, Hernández-González IA, Lopez-Arevalo P, Alavi H. On supporting traffic management decisions according to bridges structural health monitoring within a digital twin-oriented platform. Transp Res Procedia, 2025, 90: 384-391

[2] Ali R, Kang D, Suh G, Cha Y-J. Real-time multiple damage mapping using autonomous UAV and deep faster region-based neural networks for GPS-denied structures. Autom Constr, 2021, 130: 103831

[3] Aminuddin AK, Abdul Kudus S, Jamadin A, Misnan MF, Jaini ZM, Sato A (2025) Emerging trends in machine learning applications for structural health monitoring of bridges. Asian J Civil Eng 1–21. https://doi.org/10.1007/s42107-025-01511-8

[4] Amirkhani D, Allili MS, Lapointe J-F (2025) CrackSight: An Efficient Crack Segmentation Model in Varying Acquisition Ranges and Complex Backgrounds. IEEE Transactions on Automation Science and Engineering

[5] Ataei ST, Zadeh PM, Ataei S (2025) Vision-based autonomous structural damage detection using data-driven methods. arXiv preprint arXiv:2501.16662

[6] Author TS (2025) AI-Based Damage Detection in Aged Prestressed Concrete Beams: A Multimodal Learning Approach

[7] Bandi MR, Pasupuleti LN, Das T, Guchhait S. Deep learning based damage detection of concrete structures. Asian J Civ Eng, 2024, 25(7): 5197-5204

[8] Bridge design (2017a) Part 5: Concrete (AS 5100.5:2017), Standards Australia, Sydney, Australia

[9] Bridge design (2017b) Part 8: Rehabilitation and strengthening of existing bridges (AS 5100.8:2017), Standards Australia, Sydney, Australia

[10] Cha YJ, Choi W, Büyüköztürk O. Deep learning-based crack damage detection using convolutional neural networks. Computer-Aided Civil and Infrastructure Engineering, 2017, 32(5): 361-378

[11] Cha YJ, Choi W, Suh G, Mahmoudkhani S, Büyüköztürk O. Autonomous structural visual inspection using region-based deep learning for detecting multiple damage types. Computer-Aided Civil and Infrastructure Engineering, 2018, 33(9): 731-747

[12] Chen K, Lin J, You B, Luo H. Fully automatic surface defect detection of CFRP using computer vision and an augmented YOLOv8 model. J Perform Constr Facil, 2025, 39(3): 04025006

[13] Chen LC, Zhu Y, Papandreou G, Schroff F, Adam H (2018) Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation. ECCV, pp. 801–818 https://doi.org/10.1007/978-3-030-01234-2_49

[14] Concrete - Control and assessment of compressive strength (2014) TCVN 10303:2014; 0870312642; VSQI; Ha Noi, Vietnam

[15] Dang TD, Tran DT, Nguyen-Minh L, Nassif AY. Shear resistant capacity of steel fibres reinforced concrete deep beams: an experimental investigation and a new prediction model. Struct, 2021, 33: 2284-2300

[16] Design of concrete and reinforced concrete structures (2018) TCVN 5574:2018; 0870312642; VSQI; Ha Noi, Vietnam

[17] Ding Z, Fu F, Zheng J, Yang H, Zou F, Linghua K. Intelligent wood inspection approach utilizing enhanced swin transformer. IEEE Access, 2024, 12: 16794-16804

[18] Eshwar N, Ibell TJ, Nanni A. Effectiveness of CFRP strengthening on curved soffit RC beams. Adv Struct Eng, 2005, 8(1): 55-68

[19] Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening Existing Structures - Materials, RC and PC structures, masonry structures (2013) (CNR-DT 200 R1/2013); CNR; Rome

[20] Hasnat A, Islam MM, Amin AFMS. Enhancing the debonding strain limit for CFRP-strengthened RC beams using U-clamps: identification of design parameters. J Compos Constr, 2016, 20(1): 04015039

[21] Hussain T, Li Y, Ren M, Li J. Pixel-level crack segmentation and quantification enabled by multi-modality cross-fusion of RGB and depth images. Constr Build Mater, 2025, 487: 141961

[22] Jiao P, Li Q, Sun Y. Crack detection using 1D-CNN on structural accelerometer data under noise. Struct Control Health Monit, 2021, 2810e2817

[23] Pham TN, Nguyen VH, Kwon KR, Kim JH, Huh JH (2024) Improved YOLOv5 based deep learning system for jellyfish detection. IEEE Access 12, pp. 87838-87849

[24] Kang D, Benipal SS, Gopal DL, Cha Y-J. Hybrid pixel-level concrete crack segmentation and quantification across complex backgrounds using deep learning. Autom Constr, 2020, 118: 103291

[25] Katunin A, Huňady R, Hagara M. Damage identification in composite structures using high-speed 3D digital image correlation and wavelet analysis of mode shapes. Nondestruct Test Eval, 2025, 4083650-3668

[26] Kim B, Lee J. Structural health monitoring using deep convolutional neural networks with frequency response data. Smart Struct Syst, 2019, 23(4): 405-418

[27] Lou Y, Meng S, Zhou Y. Deep learning-based three-dimensional crack damage detection method using point clouds without color information. Struct Health Monit, 2025, 24(2): 657-675

[28] Lu W, Qian M, Xia Y, Lu Y, Shen J, Fu Q, Lu YElsevier. Crack _ PSTU: crack detection based on the U-Net framework combined with Swin Transformer. Structures, 2024, 62: 106241

[29] Masciotta MG, Barontini A, Chaves E, Mendes N, Brando G, Lourenço PB. Modal-based multi-criteria optimization of sensor placement techniques for dynamic monitoring of bridges. International Journal of Bridge Engineering, Management and Research, 2025, 2(2): 214250017–1-214250017–21

[30] Murphy M, Belarbi A, Bae S-W. Behavior of prestressed concrete I-girders strengthened in shear with externally bonded fiber-reinforced-polymer sheets. PCI J, 2012, 57(3): 63-82

[31] Naaman AE (1990) A New Methodology for the Analysis of Beams Prestressed With External or Unbonded Tendons. ACI Symposium Publication 120:339–354 https://doi.org/10.14359/2765

[32] Nguyen-Minh L, Phan-Vu P, Tran-Thanh D, Pham TM, Ngo-Huu C, Rovňák M. Flexural-strengthening efficiency of CFRP sheets for unbonded post-tensioned concrete T-beams. Eng Struct, 2018, 166: 1-15

[33] Nguyen-Minh L, Phan-Vu P, Tran-Thanh D, Phuong Thi Truong Q, Pham TM, Ngo-Huu C, Rovňák M. Flexural-strengthening efficiency of CFRP sheets for unbonded post-tensioned concrete T-beams. Eng Struct, 2018, 166: 1-15

[34] Nguyen-Minh L, Vo-Le D, Tran-Thanh D, Pham TM, Ho-Huu C, Rovňák M. Shear capacity of unbonded post-tensioned concrete T-beams strengthened with CFRP and GFRP U-wraps. Compos Struct, 2018, 184: 1011-1029

[35] Ni Y, Mao J, Wang H, Xi Z, Chen Z. Surface damage detection and localization for bridge visual inspection based on deep learning and 3D reconstruction. Struct Control Health Monit, 2024, 2024(1): 9988793

[36] Pan H, Yang Y, Zhang L. Bridge anomaly detection using transformer-based encoder-decoder model. Autom Constr, 2022, 137: 104209

[37] Pei X-Y, Hou Y, Huang H-B, Zheng J-X. A multi-objective sensor placement method considering modal identification uncertainty and damage detection sensitivity. Buildings, 2025, 155821

[38] Phan-Vu P, Tran DT, Pham TM, Dang TD, Ngo-Huu C, Nguyen-Minh L. Distinguished bond behaviour of CFRP sheets in unbonded post-tensioned reinforced concrete beams versus single-lap shear tests. Eng Struct, 2021, 234: 111794

[39] Pirrò M, Pereira S, Gentile C, Magalhães F, Cunha Á. Damage detection under environmental and operational variability using the cointegration technique. Eng Struct, 2025, 328: 119700

[40] Reed CE, Peterman RJ. Evaluation of prestressed concrete girders strengthened with carbon fiber reinforced polymer sheets. J Bridge Eng, 2004, 92185-192

[41] Ren S, He K, Girshick R, Sun J. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Trans Pattern Anal Mach Intell, 2017, 39(6): 1137-1149

[42] Ren Y, Bareille O, Lin Z, Huang X-R. Review of damage detection techniques in vibration-based structural health monitoring. Int J Dyn Control, 2025, 13(3): 99

[43] Ronneberger O, Fischer P, Brox T (2015) U-Net: Convolutional Networks for Biomedical Image Segmentation. Medical Image Computing and Computer-Assisted Intervention (MICCAI) pp. 234–241 https://doi.org/10.1007/978-3-319-24574-4_28

[44] Rosenboom O, Hassan TK, Rizkalla S. Flexural behavior of aged prestressed concrete girders strengthened with various FRP systems. Constr Build Mater, 2007, 214764-776

[45] Shahin M, Chen FF, Maghanaki M, Hosseinzadeh A, Zand N, Khodadadi Koodiani H. Improving the concrete crack detection process via a hybrid visual transformer algorithm. Sensors (Basel), 2024, 24(10): 3247

[46] Shen D, Jiao Y, Li M, Liu C, Wang W. Behavior of a 60-year-old reinforced concrete box beam strengthened with basalt FRP. J Adv Concr Technol, 2021, 19111100-1119

[47] Shen D, Jiao Y, Li M, Liu C, Wang W. Behavior of a 60-year-old reinforced concrete box beam strengthened with basalt fiber-reinforced polymers using steel plate anchorage. J Adv Concr Technol, 2021, 19(11): 1100-1119

[48] Song F, Liu B, Yuan G. Pixel-level crack identification for bridge concrete structures using unmanned aerial vehicle photography and deep learning. Struct Control Health Monit, 2024, 2024(1): 1299095

[49] Tran DT, Pham TM, Hao H, Chen W. Numerical study on bending response of precast segmental concrete beams externally prestressed with FRP tendons. Eng Struct, 2021, 241: 112423

[50] Tran DT, Pham TM, Hao H, Tran TT, Chen W. Impact response of prestressed prefabricated segmental and monolithic basalt-FRP-reinforced geopolymer concrete beams. J Compos Constr, 2023, 27504023045

[51] Tran DT, Pham TM, Hao H. Experimental and analytical investigations of prefabricated segmental concrete beams post-tensioned with unbonded steel/FRP tendons subjected to impact loading. Int J Impact Eng, 2024, 186: 104868

[52] Tran DT, Phan-Vu P, Nguyen-Minh L, Pham TM. Effects of pre-cracking on effectiveness of external fibre-reinforced polymer strengthening in unbonded prestressed beams. Eng Struct, 2025, 328: 119719

[53] Tran DT, Phan-Vu P, Nguyen-Minh L, Nguyen PT, Pham TM. Effects of pre-cracking on effectiveness of external fibre-reinforced polymer strengthening in unbonded prestressed beams. Eng Struct, 2025, 328: 119719

[54] Truong QPT, Phan-Vu P, Tran-Thanh D, Dang TD, Nguyen-Minh L (2018) Flexural Behavior of Unbonded Post-Tensioned Concrete T-Beams Externally Bonded With CFRP Sheets Under Static Loading," in International Conference on Advances in Computational Mechanics 2017 (ACOME 2017). Lecture Notes in Mechanical Engineering, Phu Quoc, Vietnam. Springer, pp. 273–289. https://doi.org/10.1007/978-981-10-7149-2_19

[55] Vo-Le D, Tran DT, Pham TM, Ho-Huu C, Nguyen-Minh L. Re-evaluation of shear contribution of CFRP and GFRP sheets in concrete beams post-tensioned with unbonded tendons. Eng Struct, 2022, 259: 114173

[56] Wang Y, Zhang Y, Liu Y. Impact localization on concrete slabs using GRU neural networks and time delay features. Mech Syst Signal Process, 2020, 140: 106676

[57] Waqas A, Kang D, Cha Y-J. Deep learning-based obstacle-avoiding autonomous UAVs with fiducial marker-based localization for structural health monitoring. Struct Health Monit, 2024, 232971-990

[58] Wu C, He X, Wu Y. An object detection method for catenary component images based on improved Faster R-CNN. Meas Sci Technol, 2024, 35(8): 085406

[59] Xu Y, Liu Y, Liang M. A hybrid CNN–LSTM model for detecting structural damage using dynamic response data. Eng Struct, 2021, 228: 111459

[60] Yang X, del Rey Castillo E, Zou Y, Wotherspoon L. UAV-deployed deep learning network for real-time multi-class damage detection using model quantization techniques. Autom Constr, 2024, 159: 105254

[61] Zhang L, Yang F, Zhang YD, Zhu YJ. Road crack detection using deep convolutional neural network. Image vis Comput, 2019, 89: 103-112

RIGHTS & PERMISSIONS

The Author(s)