Structural health monitoring (SHM) assessment reports are crucial for bridge maintenance and management, but their generation still relies on manual analysis, with full automation not yet achieved. This study introduces an AI-agent-enabled, automated, and intelligent framework for SHM analysis and report generation. A Python-based system was developed to process heterogeneous monitoring data with minimal human intervention, while a domain-specific multimodal large language model (MLLM), fine-tuned through low-rank adaptation, serves as the reasoning component of the AI agent to address the challenge of interpreting unstructured information. Following necessary data standardization and configuration, the framework enables automated analysis, graphical interpretation, and multilevel textual summary generation, thereby bridging the gap between raw monitoring data and engineering decision support. A representative case study on the Mingzhou Bridge was used to verify the engineering feasibility and reporting capability of the proposed framework, while further applications to 10 in-service bridges (5466 GB in total) were conducted to demonstrate its computational efficiency, scalability, and adaptability across heterogeneous SHM systems. Across these applications, the fine-tuned MLLM processed approximately 2500 figures in a total of 308 min and generated approximately 837,000 words on typical personal computer (PC) resources, demonstrating its potential as a scalable and adaptable framework for automated SHM analysis, interpretation, and report generation.
SunLShangZXiaY, et al. Review of bridge structural health monitoring aided by big data and artificial intelligence: from condition assessment to damage detection. J Struct Eng2020; 146: 22. https://doi.org/10.1061/(asce)st.1943-541x.0002535
2.
ZhangCWMousaviAAMasriSF, et al. Vibration feature extraction using signal processing techniques for structural health monitoring: a review. Mech Syst Signal Process2022; 177: 24. https://doi.org/10.1016/j.ymssp.2022.109175
3.
GongFZLeiXMXiaY. Real-time damage identification of hinge joints in multi-girder bridges using recursive least squares solution of the characteristic equation. Eng Struct2024; 300: 15. https://doi.org/10.1016/j.engstruct.2023.117147
4.
FanCCZhengYXWangBL, et al. Damage identification method for tied arch bridge suspender based on quasi-static displacement influence line. Mech Syst Signal Process2023; 200: 16. https://doi.org/10.1016/j.ymssp.2023.110518
5.
DanDHZhengWHXuZY. Research on monitoring index of transverse cooperative working performance of assembled multi-girder bridges based on displacement spectrum similarity measure. Structures2023; 48: 1322–1332. https://doi.org/10.1016/j.istruc.2023.01.023
6.
GongFZXiaYLozanoF, et al. Experimental analysis of structural nonlinear damping ratio induced by bolt joint friction. Eng Struct2024; 314: 16. https://doi.org/10.1016/j.engstruct.2024.118220
7.
XiaYGongFShangZ, et al. Damping ratio analysis of cable-stayed bridges based on attenuation response pattern extraction from monitoring data. China J Highw Transp2023; 36: 90–101.
8.
MaTWangNHanCJ, et al. Asphalt pavement residual life assessment based on SmartRock sensors. IEEE Trans Instrum Meas2023; 72: 12. https://doi.org/10.1109/tim.2023.3267372
9.
JianXDXiaYChatziE, et al. Bridge influence surface identification using a deep multilayer perceptron and computer vision techniques. Struct Health Monit2024; 23: 1606–1626. https://doi.org/10.1177/14759217231190543
PanCKomarizadehaslSXiaY. Data anomaly detection for structural health monitoring based on computer vision and transfer learning. In: IABSE symposium Tokyo 2025: environmentally friendly technologies and structures: focusing on sustainable approaches–report, Tokyo, Japan, 18–21 May 2025, pp. 329–337. International Association for Bridge and Structural Engineering (IABSE).
13.
TangZYChenZCBaoYQ, et al. Convolutional neural network-based data anomaly detection method using multiple information for structural health monitoring. Struct Control Health Monit2019; 26: 22. https://doi.org/10.1002/stc.2296
14.
XiaYWangPSunLM. Neutral axis-based health monitoring and condition assessment techniques for concrete box girder bridges. Int J Struct Stab Dyn2019; 19: 17. https://doi.org/10.1142/s0219455419400157
15.
MammeriSBarrosBConde-CarneroB, et al. From traditional damage detection methods to Physics-Informed Machine Learning in bridges: a review. Eng Struct2025; 330: 15. https://doi.org/10.1016/j.engstruct.2025.119862
16.
Di MucciVMCardellicchioARuggieriS, et al. Artificial intelligence in structural health management of existing bridges. Autom Constr2024; 167: 24. https://doi.org/10.1016/j.autcon.2024.105719
17.
XieQMaWYangW. Design and implementation of SHM system for highway tunnels. China Saf Sci J2022; 32: 56–62.
18.
LiQSHeYHZhouK, et al. Structural health monitoring for a 600m high skyscraper. Struct Des Tall Spec Build2018; 27: 22. https://doi.org/10.1002/tal.1490
ChikashuaAArchuadzeMTsintsadzeM, et al. Optical character recognition tool for georgian handwritten text recognition based on YOLOv8. Balt J Mod Comput2025; 13: 419–435. https://doi.org/10.22364/bjmc.2025.13.2.07
21.
LiuLWolterinkJMBruneC, et al. Anatomy-aided deep learning for medical image segmentation: a review. Phys Med Biol2021; 66: 11TR01. https://doi.org/10.1088/1361-6560/abfbf4
22.
HoqueMEKipliK. Deep Learning in retinal image segmentation and feature extraction: a review. Int J Online Biomed Eng2021; 17: 103–118. https://doi.org/10.3991/ijoe.v17i14.24819
Ministry of Transport of the People’s Republic of China. Technical specifications for structural monitoring of highway bridges, JT/T 1037-2022. Beijing: Ministry of Transport of the People’s Republic of China, 2022.
25.
ZhouYSunLM. Effects of high winds on a long-span sea-crossing bridge based on structural health monitoring. J Wind Eng Ind Aerodyn2018; 174: 260–268. https://doi.org/10.1016/j.jweia.2018.01.001
26.
ZhangLYSunLMShangZQ. Real-time reliability assessment based on acceleration monitoring for bridge. Sci China Technol Sci2016; 59: 1294–1304. https://doi.org/10.1007/s11431-016-6059-5
27.
MattsonSGPanditSM. Statistical moments of autoregressive model residuals for damage localisation. Mech Syst Signal Process2006; 20: 627–645. https://doi.org/10.1016/j.ymssp.2004.08.005
28.
GeurtsM. Time-series analysis–forecasting and control–GEP Box and GM Jenkins. J Mark Res1977; 14: 269–269. https://doi.org/10.2307/3150485
29.
OlivierASmythAW. Review of nonlinear filtering for SHM with an exploration of novel higher-order Kalman filtering algorithms for uncertainty quantification. J Eng Mech2017; 143: 24. https://doi.org/10.1061/(asce)em.1943-7889.0001276
30.
YuenKVKatafygiotisLS. Model updating using noisy response measurements without knowledge of the input spectrum. Earthq Eng Struct Dyn2005; 34: 167–187. https://doi.org/10.1002/eqe.415
31.
Amezquita-SanchezJPAdeliH. Signal processing techniques for vibration-based health monitoring of smart structures. Arch Comput Method Eng2016; 23: 1–15. https://doi.org/10.1007/s11831-014-9135-7
32.
YinfengDYingininLMingkuiX, et al. Analysis of earthquake ground motions using an improved Hilbert-Huang transform. Soil Dyn Earthq Eng2008; 28: 7–19. https://doi.org/10.1016/j.soildyn.2007.05.002
HuangNEShenZLongSR, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proc R Soc Lond A Math Phys Eng Sci1998; 454: 903–995.
35.
PakrashiVGhoshB. Application of S transform in structural health monitoring. In: 7th International symposium on nondestructive testing in civil engineering (NDTCE), Nantes, France30 June to 3 July 2009. https://www.ndt.net/search/docs.php3?id=7772
36.
SadeghzadehSKhatibiMM. Modal identification of single layer graphene nano sheets from ambient responses using frequency domain decomposition. Eur J Mech A-Solids2017; 65: 70–78. https://doi.org/10.1016/j.euromechsol.2017.03.009
37.
PeetersBRoeckGD. Stochastic system identification for operational modal analysis: a review. J Dyn Syst Meas Control Trans ASME2001; 123: 659–667. https://doi.org/10.1115/1.1410370
38.
LeiXMXiaYKomarizadehaslS, et al. Condition level deteriorations modeling of RC beam bridges with U-Net convolutional neural networks. Structures2022; 42: 333–342. https://doi.org/10.1016/j.istruc.2022.06.013
39.
EmadiSKomarizadehaslSXiaY. Application of intelligent low-cost accelerometers for bridge monitoring with a deep learning approach. Struct Control Health Monit2025; 2025: 21. https://doi.org/10.1155/stc/9835353
40.
CaiEJZhangYLuXZ, et al. Model-informed deep learning-based compressive sensing for resampling and recovering 2D SHM vision data. Autom Constr2024; 160: 21. https://doi.org/10.1016/j.autcon.2024.105295
41.
TanYKWangYLDengE, et al. Automatic damage detection and data completion for operational bridge safety using convolutional echo state networks. Autom Constr2024; 166: 17. https://doi.org/10.1016/j.autcon.2024.105606
42.
MaoJWangHSpencerBF. Toward data anomaly detection for automated structural health monitoring: exploiting generative adversarial nets and autoencoders. Struct Health Monit2021; 20: 1609–1626. https://doi.org/10.1177/1475921720924601
43.
AchiamJAdlerSAgarwalS, et al. Gpt-4 technical report. arXiv:2303.08774, 2023.
44.
LiuAFengBXueB, et al. Deepseek-v3 technical report. arXiv:2412.19437, 2024.
45.
TouvronHLavrilTIzacardG, et al. Llama: open and efficient foundation language models. arXiv:2302.13971, 2023.
46.
VaswaniAShazeerNParmarN, et al. Attention is all you need. In: 31st Annual conference on neural information processing systems (NIPS), Long Beach, CA, 4–9 December 2017, Advances in neural information processing systems 30, 2017. Neural Information Processing Systems.
47.
ZhaoWXZhouKLiJ, et al. A survey of large language models. arXiv:2303.18223, 2023.
48.
TeamGAnilRBorgeaudS, et al. Gemini: a family of highly capable multimodal models. arXiv:2312.11805, 2023.
49.
BaiSChenKLiuX, et al. Qwen2.5-vl technical report. arXiv:2502.13923, 2025.
50.
HurstALererAGoucherAP, et al. Gpt-4o system card. arXiv:2410.21276, 2024.
51.
ChenJZhuDShenX, et al. Minigpt-v2: large language model as a unified interface for vision-language multi-task learning. arXiv:2310.09478, 2023.
52.
LialinVDeshpandeVRumshiskyA. Scaling down to scale up: a guide to parameter-efficient fine-tuning. arXiv:2303.15647, 2023.
53.
UhmMKimJAhnS, et al. Effectiveness of retrieval augmented generation-based large language models for generating construction safety information. Autom Constr2025; 170: 25. https://doi.org/10.1016/j.autcon.2024.105926
54.
LeeJAhnSKimD, et al. Performance comparison of retrieval-augmented generation and fine-tuned large language models for construction safety management knowledge retrieval. Autom Constr2024; 168: 12. https://doi.org/10.1016/j.autcon.2024.105846
55.
RabbiABJeelaniI. AI integration in construction safety: Current state, challenges, and future opportunities in text, vision, and audio based applications. Autom Constr2024; 164: 18. https://doi.org/10.1016/j.autcon.2024.105443
56.
QinSZGuanHLiaoWJ, et al. Intelligent design and optimization system for shear wall structures based on large language models and generative artificial intelligence. J Build Eng2024; 95: 18. https://doi.org/10.1016/j.jobe.2024.109996
57.
WangXYueQRLiuXG. Crack image classification and information extraction in steel bridges using multimodal large language models. Autom Constr2025; 171: 17. https://doi.org/10.1016/j.autcon.2025.105995
58.
ZhangCHLeiXMXiaY, et al. Automatic bridge inspection database construction through hybrid information extraction and large language models. Dev Built Environ2024; 20: 16. https://doi.org/10.1016/j.dibe.2024.100549
59.
JianXDZhongHQXiaY, et al. Faulty data detection and classification for bridge structural health monitoring via statistical and deep-learning approach. Struct Control Health Monit2021; 28: 20. https://doi.org/10.1002/stc.2824
60.
RosnerB. Percentage points for a generalized ESD many-outlier procedure. Technometrics1983; 25: 165–172.
61.
MoorBDVanPSuykensOJ. Subspace algorithms for system identification and stochastic realization. In: Proceedings of the international symposium on recent advances in mathematical theory of systems, control, networks and signal processing (MTNS’91), Kobe, Japan, 1991, pp. 589–595.
62.
CampelloRJMoulaviDSanderJ. Density-based clustering based on hierarchical density estimates. In: Pacific-Asia conference on knowledge discovery and data mining, (PAKDD 2013), Gold Coast, QLD, Australia, 14–17 April 2013, pp. 160–172. Springer.
63.
FarrarCRWordenK. Structural health monitoring: a machine learning perspective. John Wiley & Sons, 2012.
64.
GururanganSMarasovićASwayamdiptaS, et al. Don’t stop pretraining: adapt language models to domains and tasks. arXiv:2004.10964, 2020.
65.
ComaniciGBieberESchaekermannM, et al. Gemini 2.5: pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arXiv:2507.06261, 2025.
66.
ReynoldsLMcDonellK. Prompt programming for large language models: beyond the few-shot paradigm. In: Extended abstracts of the 2021 CHI conference on human factors in computing systems, Yokohama, Japan, 8–13 May 2021, pp. 1–7. Association for Computing Machinery (ACM).
67.
WeiJWangXSchuurmansD, et al. Chain-of-thought prompting elicits reasoning in large language models. Adv Neural Inform Process Syst2022; 35: 24824–24837.
68.
WangXWeiJSchuurmansD, et al. Self-consistency improves chain of thought reasoning in language models. arXiv:2203.11171, 2022.
69.
HuEJShenYWallisP, et al. Lora: Low-rank adaptation of large language models. ICLR2022; 1: 3.
70.
PapineniKRoukosSWardT, et al. Bleu: a method for automatic evaluation of machine translation. In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics, Philadelphia, PA, USA, 7–12 July 2002, pp. 311–318. Association for Computational Linguistics.
71.
LinC-Y. Rouge: a package for automatic evaluation of summaries. In: Text summarization branches out, Barcelona, Spain, 25–26 July 2004, pp. 74–81. Association for Computational Linguistics.
72.
HongWYuWGuX, et al. Glm-4.1 v-thinking: towards versatile multimodal reasoning with scalable reinforcement learning. arXiv: 2507.01006, 2025.
73.
PeetersBDe RoeckG. One-year monitoring of the Z24-Bridge: environmental effects versus damage events. Earthq Eng Struct Dyn2001; 30: 149–171.
LuoLSongMZhongH, et al. Hierarchical Bayesian model updating of a long-span arch bridge considering temperature and traffic loads. Mech Syst Signal Process2024; 210: 111152. https://doi.org/10.1016/j.ymssp.2024.111152
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
