pith. sign in

arxiv: 2605.12916 · v1 · pith:F3BCRY76new · submitted 2026-05-13 · 💻 cs.MA · cs.LG

SHM-Agents: A Generalist-Specialist Integrated Agent System for Structural Health Monitoring

Yuequan Bao , Xing Li , Huabin Sun , Dawei Liu , Yuxuan Tian , Haiyang Hu This is my paper

Pith reviewed 2026-06-30 21:47 UTC · model grok-4.3

classification 💻 cs.MA cs.LG
keywords structural health monitoringagent systemlarge language modelsbridge monitoringmodal identificationdamage identificationfatigue estimationreliability assessment
0
0 comments X

The pith

SHM-Agents combines large language models with specialized algorithms so engineers can run structural health monitoring tasks through natural language.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes SHM-Agents as a system that pairs the planning and reasoning of large language models with existing specialized algorithms for structural health monitoring. This integration aims to remove high implementation barriers, limited interoperability, and complex training that currently restrict use of those algorithms. Experiments on a long-span cable-stayed bridge show the system handling twelve tasks from anomaly diagnosis through fatigue estimation and knowledge questions, all triggered by plain-language requests. If the approach works, monitoring workflows become accessible without custom coding for each new job or deep expertise in every algorithm. The modular design is presented as the route for adding further capabilities over time.

Core claim

SHM-Agents is a generalist-specialist agent system that integrates the reasoning and planning abilities of large language models with the problem-solving strengths of specialized algorithms, enabling end-to-end execution of single and combined SHM tasks via natural language and supporting deep learning pre-training and modular expansion, with experiments demonstrating accurate and efficient performance across data anomaly diagnosis and recovery, signal processing, statistical analysis, modal identification, damage identification, finite element model updating, vehicle load modeling, response calculation, reliability assessment, fatigue estimation and bridge knowledge Q&A on a long-span cable

What carries the argument

The SHM-Agents generalist-specialist agent system, which wraps specialized SHM algorithms for invocation by an LLM-based planner.

If this is right

  • Users can execute diverse SHM tasks end-to-end through natural language instructions.
  • Deployment is simplified by support for deep learning pre-training.
  • New capabilities can be added through the modular design without rebuilding the system.
  • A single platform handles the full range from data anomaly diagnosis to reliability assessment and fatigue estimation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same wrapping pattern could be tested on monitoring problems outside civil engineering where specialized tools already exist.
  • Real-time conversational dashboards might emerge if the planner can maintain state across sequential tasks on streaming sensor data.
  • Standardized wrapper interfaces for legacy SHM code would reduce the engineering effort needed to adopt the approach.

Load-bearing premise

Specialized SHM algorithms can be wrapped and invoked reliably by the LLM planner without introducing new implementation barriers or accuracy losses.

What would settle it

A side-by-side test on the same bridge dataset where the agent-invoked versions of the specialized algorithms produce measurably lower accuracy or higher error rates than direct execution of those algorithms.

Figures

Figures reproduced from arXiv: 2605.12916 by Dawei Liu, Haiyang Hu, Huabin Sun, Xing Li, Yuequan Bao, Yuxuan Tian.

Figure 1
Figure 1. Figure 1: To streamline agent development, researchers have introduced numerous open-source frameworks (Jain (2022), Qian et al. (2024); Alon et al. (2019); Duan and Wang (2024); Hong et al. (2024); Wang et al. (2024); Shinn et al. (2023); Lozhkov et al. (2024); Schick et al. (2023); Webb et al. (2025)), including ChatDev (Qian et al. (2024)), Auto￾GPT (Alon et al. (2019)), MetaGPT (Hong et al. (2024)), LangChain (W… view at source ↗
Figure 2
Figure 2. Figure 2: The functions of the SHM-Agents. Mendible-Barreto et al. (2025) proposed a modular multi-agent LLM framework for scientific tasks, enabling the collaborative execution of complex workflows through the flexible integration of customized Python functions. Building on this framework, they developed DynaMate, a system that automates the generation, execution and analysis of molecular simulations. Wang et al. (… view at source ↗
Figure 3
Figure 3. Figure 3: The Architecture of SHM-Agents. SHM-Agents comprise three key components: process agent nodes, skill agent nodes and global modules. The process agent nodes oversee task planning, allocation, summarization and output. The skill agent nodes are tasked with executing specific SHM algorithms. The global modules manages operational information, including memory, data and configurations. Conditional edges refer… view at source ↗
Figure 4
Figure 4. Figure 4: The flowchart of SHM-Agents executing specific tasks. the workflow proceeds to the next task step. If an error cannot be corrected, the Allocate node reports it directly to the user. Once all subtasks have been completed, the Allocate node terminates the execution process and transfers control to the Summary node for final output. This distinctive “allocate￾execute-feedback-reallocate” mechanism enables au… view at source ↗
Figure 5
Figure 5. Figure 5: The user interface and Instructions of SHM-Agents. x z Monitoring point. WIM. y z 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Diagram of the structure of a cable-stayed bridge. structural diagram of this bridge. It is a dual-direction, six￾lane highway bridge, designed to accommodate a Highway Level I vehicle load. Eleven acceleration sensors are installed along the y-axis of the bridge, which collect acceleration data at a frequency of 50Hz. Moreover, a WIM system is installed to gather information on passing vehicles. Configura… view at source ↗
Figure 7
Figure 7. Figure 7: The finite element model of the cable-stayed bridge. 4. Bridge information: In this case, bridge information, encompassing construction details and maintenance inspection reports, is incorporated into the SHM￾Agents configuration in PDF format. Preprocessing After uploading the configuration param￾eters, the configuration generation module will perform preprocessing, including finite element model processi… view at source ↗
Figure 8
Figure 8. Figure 8: Anomaly detection results for original data. In this image, the color green indicates normal data, whereas the other colors correspond to various types of anomalies. Normal Missing Outliers Sub-minimum Trend Out-of-range Constant Drift Mutant Normal Missing Outliers Sub-minimum Trend Out-of-range Constant Drift Mutant [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Anomaly detection results for reconstructed data [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Original anomaly data and its data recovery results. (1). Apply both the Fourier transform and the wavelet transform to the input signal, and extract only the 0–1 Hz frequency components for plotting. (2). Use a single canvas with left and right subplots, with a total canvas width of 7.06. The left subplot Prepared using sagej.cls [PITH_FULL_IMAGE:figures/full_fig_p009_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Wavelet transform result. should show a 3D time-frequency plot of the wavelet transform. (3). The right side should contain: - a 2D time-frequency plot of the wavelet transform, - the original data plot, - and the Fourier transform spectrum plot. The original data plot should be placed directly above the 2D time-frequency plot, with their time axes aligned. The original data plot should have no time￾axis … view at source ↗
Figure 12
Figure 12. Figure 12: Temperature probability distribution fitting graph. the generated figure to a complete local path for proper insertion, and return the complete saved image path. Other node: The task has been completed and the result is shown in [PITH_FULL_IMAGE:figures/full_fig_p011_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Data Anomaly Diagnosis Result. The blue lines represent the time-domain signals, while the green lines correspond to the frequency-domain signals. Above each plot, the identification results of data anomaly diagnosis are indicated. (a) 1st-order mode shape (b) 2nd-order mode shape (c) 3rd-order mode shape 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 [PITH_FULL_IMAGE:figures/ful… view at source ↗
Figure 14
Figure 14. Figure 14: Mode shapes identification results. Step 2: Use the Modal identification node to identify the mode of the structure based on the input monitoring data monitoring data. 2. Allocate node: Please use the Data Anomaly Diagnosis node agent to conduct data anomaly diagnosis on the input monitoring data monitoring data. 3. Data Anomaly Diagnosis node: The data anomaly diagnosis has been successfully executed. Th… view at source ↗
Figure 15
Figure 15. Figure 15: Graph of changes in structural modal parameters. modal identification. Initially, the data anomaly detection algorithm is used to identify abnormalities. As shown in [PITH_FULL_IMAGE:figures/full_fig_p013_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Reliability calculation process. Monitoring data: Monitoring data from acceleration sensors. Thinking: 1. Architecture node: Step 1: Use the Reliability calculation node to calculate the structural reliability under the uncertainty file using the updated finite element model. 2. Allocate node: Please use the Reliability calculation node agent to calculate the structural reliability under the uncertainty f… view at source ↗
Figure 17
Figure 17. Figure 17: The calculation curve of structural failure probability for future years [PITH_FULL_IMAGE:figures/full_fig_p015_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: The change in structural failure probability over the next 10 years, with a time interval of 2 years [PITH_FULL_IMAGE:figures/full_fig_p015_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Fatigue damage diagram under 30-minute vehicle loading. The vehicle load with a duration of 30 minutes has been successfully generated. The results are stored in a Python dictionary named V load 4. Allocate node: Please use the Fatigue calculation node agent to calculate the structural fatigue damage under the generated vehicle load V load for half an hour. 5. Fatigue calculation node: The fatigue damage … view at source ↗
Figure 20
Figure 20. Figure 20: Mid-span displacement diagram under 10s vehicle loading. Please calculate the mid-span displacement of the structure under a 10-second vehicle load. Thinking: 1. Architecture node: Step 1: Use the Vehicle load modeling node to generate the vehicle load with a duration of 10s. Step 2: Use the Response calculation node to calculate the structural response under the generated vehicle load. 2.Allocate node: P… view at source ↗
read the original abstract

Artificial intelligence is increasingly used to simplify complex tasks. In engineering applications of structural health monitoring (SHM), existing specialized algorithms, while effective, often face high implementation barriers, limited interoperability and complex training procedures. To overcome these challenges, this paper proposes SHM-Agents, a generalist-specialist agent system that integrates the reasoning and planning abilities of large language models with the problem-solving strengths of specialized algorithms. SHM-Agents enables end-to-end execution of single and combined SHM tasks via natural language, supports deep learning pre-training to simplify deployment and allows flexible expansion through a modular design. Experiments on a long-span cable-stayed bridge show that SHM-Agents can accurately and efficiently perform diverse SHM tasks, including data anomaly diagnosis and recovery, signal processing, statistical analysis, modal identification, damage identification, finite element model updating, vehicle load modeling, response calculation, reliability assessment, fatigue estimation and bridge knowledge Q\&A.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The manuscript proposes SHM-Agents, a generalist-specialist agent system integrating large language models for reasoning and planning with specialized structural health monitoring (SHM) algorithms. It enables end-to-end natural-language execution of single and combined SHM tasks, supports modular expansion and deep learning pre-training for simplified deployment, and is evaluated on a long-span cable-stayed bridge for tasks including data anomaly diagnosis/recovery, signal processing, modal identification, damage identification, finite element model updating, vehicle load modeling, response calculation, reliability assessment, fatigue estimation, and bridge knowledge Q&A.

Significance. If the results hold, the approach could meaningfully reduce implementation barriers and interoperability issues in SHM by allowing natural-language orchestration of proven specialized tools without new accuracy losses. The modular wrapper design and reported task success rates on a real bridge dataset are strengths; the work also demonstrates practical multi-task capability across diagnosis, identification, updating, and assessment stages.

minor comments (3)
  1. [Abstract] Abstract: the claim of 'accurate and efficient' performance would be strengthened by including one or two key quantitative metrics (e.g., success rates or error measures) from the cable-stayed bridge experiments.
  2. [Experiments] Experiments section: while task success rates are reported, adding a brief comparison against direct invocation of the same specialized algorithms (without the LLM planner) would clarify whether the integration introduces any overhead.
  3. [Methodology] Methodology: the description of the modular wrappers would benefit from an explicit data-flow diagram or pseudocode showing how the LLM planner selects and invokes the specialist modules.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary of SHM-Agents, the recognition of its practical contributions to reducing implementation barriers in SHM, and the recommendation for minor revision. No specific major comments were provided in the report.

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper describes a modular agent system (SHM-Agents) that wraps existing specialized SHM algorithms under LLM planning and reports experimental success rates on a cable-stayed bridge. No derivation chain, equations, fitted parameters, or self-citation load-bearing steps are present; the claims rest on empirical task performance rather than any reduction of outputs to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

The contribution rests on the unexamined premise that LLM planning plus existing SHM toolkits can be composed without loss of accuracy or added complexity; no free parameters, axioms, or invented physical entities are stated.

invented entities (1)
  • SHM-Agents no independent evidence
    purpose: Generalist-specialist agent system for end-to-end SHM via natural language
    The paper introduces this named system as its central artifact.

pith-pipeline@v0.9.1-grok · 5708 in / 1121 out tokens · 22664 ms · 2026-06-30T21:47:54.057406+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

55 extracted references · 51 canonical work pages · 5 internal anchors

  1. [1]

    Measurement 129: 589--606

    Alavi AH, Jiao P, Buttlar WG and Lajnef N (2018) Internet of things-enabled smart cities: State-of-the-art and future trends. Measurement 129: 589--606. doi:10.1016/j.measurement.2018.07.067

    work page doi:10.1016/j.measurement.2018.07.067 2018

  2. [2]

    Code2vec: Learning distributed representations of code,

    Alon U, Zilberstein M, Levy O and Yahav E (2019) code2vec: Learning distributed representations of code. Proceedings of the ACM on Programming Languages 3(POPL): 1--29. doi:10.1145/3290353

    work page doi:10.1145/3290353 2019

  3. [3]

    Procedia Computer Science 246: 3781--3790

    Arslan M, Ghanem H, Munawar S and Cruz C (2024) A survey on rag with llms. Procedia Computer Science 246: 3781--3790. doi:10.1016/j.procs.2024.09.178

    work page doi:10.1016/j.procs.2024.09.178 2024

  4. [4]

    Journal of Sound and Vibration 333(23): 6057--6073

    Asgarieh E, Moaveni B and Stavridis A (2014) Nonlinear finite element model updating of an infilled frame based on identified time-varying modal parameters during an earthquake. Journal of Sound and Vibration 333(23): 6057--6073. doi:10.1016/j.jsv.2014.04.064

    work page doi:10.1016/j.jsv.2014.04.064 2014

  5. [5]

    Engineering 5(2): 234--242

    Bao Y, Chen Z, Wei S, Xu Y, Tang Z and Li H (2019) The state of the art of data science and engineering in structural health monitoring. Engineering 5(2): 234--242. doi:10.1016/j.eng.2018.11.027

    work page doi:10.1016/j.eng.2018.11.027 2019

  6. [6]

    Structural Health Monitoring 19(6): 1963--1975

    Bao Y, Guo Y and Li H (2020) A machine learning--based approach for adaptive sparse time--frequency analysis used in structural health monitoring. Structural Health Monitoring 19(6): 1963--1975. doi:10.1177/1475921720909440

    work page doi:10.1177/1475921720909440 2020

  7. [7]

    Mechanical Systems and Signal Processing 216: 111458

    Bao Y, Liu D and Li H (2024) A mechanics-informed neural network method for structural modal identification. Mechanical Systems and Signal Processing 216: 111458. doi:10.1016/j.ymssp.2024.111458

    work page doi:10.1016/j.ymssp.2024.111458 2024

  8. [8]

    doi:10.2139/ssrn.5099636

    Bao Y and Tian Y (2025) A graph neural network-based method for orthotropic steel deck fatigue damage assessment due to vehicular loading. doi:10.2139/ssrn.5099636. SSRN

    work page doi:10.2139/ssrn.5099636 2025

  9. [9]

    Nature Communications 12(1): 6136

    Chen Z, Liu Y and Sun H (2021) Physics-informed learning of governing equations from scarce data. Nature Communications 12(1): 6136. doi:10.1038/s41467-021-26434-1

    work page doi:10.1038/s41467-021-26434-1 2021

  10. [10]

    DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model

    DeepSeek-AI et al. (2024) DeepSeek-V2 : A strong, economical, and efficient mixture-of-experts language model. doi:10.48550/arXiv.2405.04434

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2405.04434 2024

  11. [11]

    Engineering Structures 178: 309--317

    Deng L, Yan W and Nie L (2019) A simple corrosion fatigue design method for bridges considering the coupled corrosion-overloading effect. Engineering Structures 178: 309--317. doi:10.1016/j.engstruct.2018.10.028

    work page doi:10.1016/j.engstruct.2018.10.028 2019

  12. [12]

    doi:10.48550/arXiv.2411.18241

    Duan Z and Wang J (2024) Exploration of llm multi-agent application implementation based on langgraph+crewai. doi:10.48550/arXiv.2411.18241

    work page doi:10.48550/arxiv.2411.18241 2024

  13. [13]

    doi:10.48550/arXiv.2402.06664

    Fang R, Bindu R, Gupta A, Zhan Q and Kang D (2024) Llm agents can autonomously hack websites. doi:10.48550/arXiv.2402.06664

    work page doi:10.48550/arxiv.2402.06664 2024

  14. [14]

    Gemini: A Family of Highly Capable Multimodal Models

    Gemini Team et al. (2025) Gemini: A family of highly capable multimodal models. doi:10.48550/arXiv.2312.11805

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2312.11805 2025

  15. [15]

    (2025) From llms to agents: A comparative evaluation of llms and llm-based agents in security patch detection

    Han J et al. (2025) From llms to agents: A comparative evaluation of llms and llm-based agents in security patch detection. doi:10.48550/arXiv.2511.08060

    work page doi:10.48550/arxiv.2511.08060 2025

  16. [16]

    Micromachines 12(7): 813

    Han Z, Jiao P and Zhu Z (2021) Combination of piezoelectric and triboelectric devices for robotic self-powered sensors. Micromachines 12(7): 813. doi:10.3390/mi12070813

    work page doi:10.3390/mi12070813 2021

  17. [17]

    Automation in Construction 136: 104168

    He Z, Li W, Salehi H, Zhang H, Zhou H and Jiao P (2022) Integrated structural health monitoring in bridge engineering. Automation in Construction 136: 104168. doi:10.1016/j.autcon.2022.104168

    work page doi:10.1016/j.autcon.2022.104168 2022

  18. [18]

    MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework

    Hong S et al. (2024) Metagpt: Meta programming for a multi-agent collaborative framework. doi:10.48550/arXiv.2308.00352

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2308.00352 2024

  19. [19]

    Structural Health Monitoring : 14759217251395197doi:10.1177/14759217251395197

    Hou R, Che L and Bao Y (2025) Accelerating bayesian inference of structural damage detection based on physics-guided cnn and zero-shot transfer learning. Structural Health Monitoring : 14759217251395197doi:10.1177/14759217251395197

    work page doi:10.1177/14759217251395197 2025

  20. [20]

    In: Introduction to Transformers for NLP

    Jain SM (2022) Hugging face. In: Introduction to Transformers for NLP. Berkeley, CA: Apress, pp. 51--67. doi:10.1007/978-1-4842-8844-3_4

    work page doi:10.1007/978-1-4842-8844-3_4 2022

  21. [21]

    International Journal of Remote Sensing 42(14): 5277--5316

    Jiao P, Wang J, Chen X, Ruan J, Ye X and Alavi AH (2021) Next-generation remote sensing and prediction of sand and dust storms: State-of-the-art and future trends. International Journal of Remote Sensing 42(14): 5277--5316. doi:10.1080/01431161.2021.1912433

    work page doi:10.1080/01431161.2021.1912433 2021

  22. [22]

    (2025) Agent4vul: multimodal llm agents for smart contract vulnerability detection

    Jie W et al. (2025) Agent4vul: multimodal llm agents for smart contract vulnerability detection. Science China Information Sciences 68(6): 160101. doi:10.1007/s11432-024-4402-2

    work page doi:10.1007/s11432-024-4402-2 2025

  23. [23]

    In: Proceedings of the 40th ACM/SIGAPP Symposium on Applied Computing

    Kim J, Hur M and Min M (2025) From rag to qa-rag: Integrating generative ai for pharmaceutical regulatory compliance process. In: Proceedings of the 40th ACM/SIGAPP Symposium on Applied Computing. Catania International Airport, Catania, Italy: ACM, pp. 1293--1295. doi:10.1145/3672608.3707749

    work page doi:10.1145/3672608.3707749 2025

  24. [24]

    Engineering, Technology & Applied Science Research 14(3): 14219--14224

    Kumar A, Sharma R and Bedi P (2024) Towards optimal nlp solutions: Analyzing gpt and llama-2 models across model scale, dataset size, and task diversity. Engineering, Technology & Applied Science Research 14(3): 14219--14224. doi:10.48084/etasr.7200

    work page doi:10.48084/etasr.7200 2024

  25. [25]

    Computer Communications 160: 71--80

    Li D, Cai Z, Qin B and Deng L (2020) Signal frequency domain analysis and sensor fault diagnosis based on artificial intelligence. Computer Communications 160: 71--80. doi:10.1016/j.comcom.2020.05.034

    work page doi:10.1016/j.comcom.2020.05.034 2020

  26. [26]

    Reliability Engineering & System Safety 243: 109898

    Li W, Geng R and Chen S (2024) Csp-free adaptive kriging surrogate model method for reliability analysis with small failure probability. Reliability Engineering & System Safety 243: 109898. doi:10.1016/j.ress.2023.109898

    work page doi:10.1016/j.ress.2023.109898 2024

  27. [27]

    StarCoder 2 and The Stack v2: The Next Generation

    Lozhkov A et al. (2024) Starcoder 2 and the stack v2: The next generation. doi:10.48550/arXiv.2402.19173

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2402.19173 2024

  28. [28]

    Journal of Traffic and Transportation Engineering (English Edition) 2(4): 242--248

    Mei K, Li Y and Lu Z (2015) Application study on the first cable-stayed bridge with cfrp cables in china. Journal of Traffic and Transportation Engineering (English Edition) 2(4): 242--248. doi:10.1016/j.jtte.2015.05.004

    work page doi:10.1016/j.jtte.2015.05.004 2015

  29. [29]

    Molecular Systems Design & Engineering 10(7): 585--598

    Mendible-Barreto OA, Díaz-Maldonado M, Carmona Esteva FJ, Torres JE, Córdova-Figueroa UM and Colón YJ (2025) Dynamate: leveraging ai-agents for customized research workflows. Molecular Systems Design & Engineering 10(7): 585--598. doi:10.1039/D5ME00062A

    work page doi:10.1039/d5me00062a 2025

  30. [30]

    GPT-4 Technical Report

    OpenAI et al. (2024) GPT-4 technical report. doi:10.48550/arXiv.2303.08774

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.08774 2024

  31. [31]

    Structural Health Monitoring : 14759217251322943doi:10.1177/14759217251322943

    Pan Q and Bao Y (2025) A hybrid attention hierarchical network-based extreme event detection method for structural health monitoring. Structural Health Monitoring : 14759217251322943doi:10.1177/14759217251322943

    work page doi:10.1177/14759217251322943 2025

  32. [32]

    Structural Health Monitoring 22(5): 3077--3091

    Pan Q, Bao Y and Li H (2023) Transfer learning-based data anomaly detection for structural health monitoring. Structural Health Monitoring 22(5): 3077--3091. doi:10.1177/14759217221142174

    work page doi:10.1177/14759217221142174 2023

  33. [33]

    doi:10.48550/arXiv.2511.03878

    Prasai S, Du M, Zhang Y and Yang F (2025) Knowthyself: An agentic assistant for llm interpretability. doi:10.48550/arXiv.2511.03878

    work page doi:10.48550/arxiv.2511.03878 2025

  34. [34]

    (2024) Chatdev: Communicative agents for software development

    Qian C et al. (2024) Chatdev: Communicative agents for software development. In: Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Bangkok, Thailand: Association for Computational Linguistics, pp. 15174--15186. doi:10.18653/v1/2024.acl-long.810

    work page doi:10.18653/v1/2024.acl-long.810 2024

  35. [35]

    Engineering Structures 40: 413--435

    Ribeiro D, Calçada R, Delgado R, Brehm M and Zabel V (2012) Finite element model updating of a bowstring-arch railway bridge based on experimental modal parameters. Engineering Structures 40: 413--435. doi:10.1016/j.engstruct.2012.03.013

    work page doi:10.1016/j.engstruct.2012.03.013 2012

  36. [36]

    Engineering Structures 102: 66--79

    Sanayei M, Khaloo A, Gul M and Catbas FN (2015) Automated finite element model updating of a scale bridge model using measured static and modal test data. Engineering Structures 102: 66--79. doi:10.1016/j.engstruct.2015.07.029

    work page doi:10.1016/j.engstruct.2015.07.029 2015

  37. [37]

    (2023) Toolformer: Language models can teach themselves to use tools

    Schick T et al. (2023) Toolformer: Language models can teach themselves to use tools. Advances in Neural Information Processing Systems 36: 68539--68551

    2023

  38. [38]

    Advances in Neural Information Processing Systems 36: 8634--8652

    Shinn N, Cassano F, Gopinath A, Narasimhan K and Yao S (2023) Reflexion: Language agents with verbal reinforcement learning. Advances in Neural Information Processing Systems 36: 8634--8652

    2023

  39. [39]

    Reliability Engineering & System Safety 262: 111169

    Sun H and Bao Y (2025) An active learning method based on monte carlo dropout neural network for high-dimensional reliability analysis. Reliability Engineering & System Safety 262: 111169. doi:10.1016/j.ress.2025.111169

    work page doi:10.1016/j.ress.2025.111169 2025

  40. [40]

    Structure and Infrastructure Engineering 21(5): 742--754

    Sun H, Guan X, Bao Y, Xu Y and Li H (2025 a ) Normalising flow-based vehicle load modeling for bridges. Structure and Infrastructure Engineering 21(5): 742--754. doi:10.1080/15732479.2023.2236600

    work page doi:10.1080/15732479.2023.2236600 2025

  41. [41]

    (2025 b ) Llm agent meets agentic ai: Can llm agents simulate customers to evaluate agentic-ai-based shopping assistants? doi:10.48550/arXiv.2509.21501

    Sun L et al. (2025 b ) Llm agent meets agentic ai: Can llm agents simulate customers to evaluate agentic-ai-based shopping assistants? doi:10.48550/arXiv.2509.21501

    work page doi:10.48550/arxiv.2509.21501 2025

  42. [42]

    (2024) Bring game characters to the social space: Developing storytelling community ai agents driven by llms

    Sun Y et al. (2024) Bring game characters to the social space: Developing storytelling community ai agents driven by llms. doi:10.2139/ssrn.4806067. SSRN

    work page doi:10.2139/ssrn.4806067 2024

  43. [43]

    (2025) Llm agent for hyper-parameter optimization

    Wang W et al. (2025) Llm agent for hyper-parameter optimization. doi:10.48550/arXiv.2506.15167

    work page doi:10.48550/arxiv.2506.15167 2025

  44. [44]

    Robotics and Computer-Integrated Manufacturing 97: 103120

    Wang Y, Guo Q, Zheng L, Wang B, Zheng P and Qi Z (2026) Llm based autonomous agent of human-robot collaboration for aerospace wire harnessing assembly. Robotics and Computer-Integrated Manufacturing 97: 103120. doi:10.1016/j.rcim.2025.103120

    work page doi:10.1016/j.rcim.2025.103120 2026

  45. [45]

    doi:10.48550/arXiv.2406.18122

    Wang Z, Liu J, Zhang S and Yang Y (2024) Poisoned langchain: Jailbreak llms by langchain. doi:10.48550/arXiv.2406.18122

    work page doi:10.48550/arxiv.2406.18122 2024

  46. [46]

    Nature Communications 16(1): 8633

    Webb T, Mondal SS and Momennejad I (2025) A brain-inspired agentic architecture to improve planning with llms. Nature Communications 16(1): 8633. doi:10.1038/s41467-025-63804-5

    work page doi:10.1038/s41467-025-63804-5 2025

  47. [47]

    ACM Transactions on Software Engineering and Methodology 34(7): 1--42

    Wu JJ and Fard FH (2025) Humanevalcomm: Benchmarking the communication competence of code generation for llms and llm agents. ACM Transactions on Software Engineering and Methodology 34(7): 1--42. doi:10.1145/3715109

    work page doi:10.1145/3715109 2025

  48. [48]

    doi:10.48550/ARXIV.2402.13178

    Xiong G, Jin Q, Lu Z and Zhang A (2024) Benchmarking retrieval-augmented generation for medicine. doi:10.48550/ARXIV.2402.13178

    work page doi:10.48550/arxiv.2402.13178 2024

  49. [49]

    Structural Health Monitoring 18(3): 653--674

    Xu Y, Bao Y, Chen J, Zuo W and Li H (2019 a ) Surface fatigue crack identification in steel box girder of bridges by a deep fusion convolutional neural network based on consumer-grade camera images. Structural Health Monitoring 18(3): 653--674. doi:10.1177/1475921718764873

    work page doi:10.1177/1475921718764873 2019

  50. [50]

    Structural Control and Health Monitoring 26(3): e2313

    Xu Y, Wei S, Bao Y and Li H (2019 b ) Automatic seismic damage identification of reinforced concrete columns from images by a region-based deep convolutional neural network. Structural Control and Health Monitoring 26(3): e2313. doi:10.1002/stc.2313

    work page doi:10.1002/stc.2313 2019

  51. [51]

    Smart Structures and Systems 12(3\_4): 363--379

    Ye XW, Ni YQ, Wai TT, Wong KY, Zhang XM and Xu F (2013) A vision-based system for dynamic displacement measurement of long-span bridges: Algorithm and verification. Smart Structures and Systems 12(3\_4): 363--379. doi:10.12989/SSS.2013.12.3\_4.363

    work page doi:10.12989/sss.2013.12.3 2013

  52. [52]

    Engineering Structures 45: 166--176

    Ye XW, Ni YQ, Wong KY and Ko JM (2012) Statistical analysis of stress spectra for fatigue life assessment of steel bridges with structural health monitoring data. Engineering Structures 45: 166--176. doi:10.1016/j.engstruct.2012.06.016

    work page doi:10.1016/j.engstruct.2012.06.016 2012

  53. [53]

    doi:10.48550/ARXIV.2402.05131

    Yepes AJ, You Y, Milczek J, Laverde S and Li R (2024) Financial report chunking for effective retrieval augmented generation. doi:10.48550/ARXIV.2402.05131

    work page doi:10.48550/arxiv.2402.05131 2024

  54. [54]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution isbn journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 'mid...

  55. [55]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize ":" * " " *...