Open research questions in Structural Health Monitoring Techniques

What the literature leaves open

• Future work will focus on addressing the current limitations by incorporating variations in vehicle and bridge parameters, utilizing natural vertical ground motion components, and integrating physical constraints into the GAN framework to ensure that the generated samples better adhere to the underlying physical laws of the VTB system.

  A new interpretable dynamic ensemble learning model with sample augmentation for predicting seismic responses of vehicle-track-bridge systems · 2026 · DOI

• Future research should focus on integrating finite element model updating, wireless structural health monitoring systems, artificial intelligence, digital twins, and machine learning algorithms to improve the accuracy and efficiency of bridge assessment methodologies. Although recent studies have demonstrated promising results, further research is required to establish standardized procedures for bridge load-carrying capacity estimation using dynamic characteristics. While several countries have adopted dynamic testing as a practical alternative to static testing, comprehensive investigations under Indian bridge conditions are still lacking.

  STATE-OF-THE-ART REVIEW ON LOAD-CARRYING CAPACITY ASSESSMENT OF BRIDGES USING STATIC AND DYNAMIC METHODS · 2026 · DOI

• DIRECTIONS 10.1 Current Limitations Tower-Core v1.0.0 provides a validated framework for the continuous structural integrity assessment of vertical tower systems within several boundary conditions that define the current scope. First, the DFMM modal identification algorithm assumes stationary ambient vibration excitation from broadband wind loading. During severe storm events, the non-stationarity of the excitation may degrade the quality of frequency identification, and adaptive windowing techniques for nonstationary conditions are not yet implemented in the current module. Second, the Palmgren–Miner fatigue damage rule implemented in the SJFAM is a linear accumulation model that does not explicitly account for load sequence effects — the accelerated crack propagation that can occur when high-amplitude load cycles follow low-amplitude cycles and retardation effects that can occur in the reversed order. For tower structural details subject to rare high-amplitude loading events (extreme storms), these sequence effects may be relevant. Third, the spatial resolution of the damage TOWER-SAFETY-01 | Version 1.0.0 | MIT License | DOI: 10.5281/zenodo.20394041 github.com/gitdeeper12/TOWER-CORE TOWER-CORE — Structural Integrity Assessment and Safety Governance for Vertical Tower Systems Baladi, S. (2026) assessment is limited by the sensor spacing: the current sensor complement provides damage assessment at instrumented detail locations, with extrapolation to uninstrumented locations via the mode shape-based interpolation functions. Cracks or degradation at locations between sensors may not be detected until they have grown to a size that affects the measured modal parameters. 10.2 Future Research Directions Four priority extensions are planned for subsequent Tower-Core development. First, non-stationary modal identification: extension of the DFMM frequency tracking algorithm to non-stationary conditions through the implementation of the Short-Time SSI algorithm, which processes windowed segments of the acceleration time series and tracks the evolution of modal parameters through the loading event. This extension is particularly relevant for the identification of natural frequency shifts during severe storm events — precisely the conditions under which the overturning stability margin is most critical. Second, probabilistic fatigue life prediction: integration of a probabilistic framework for fatigue life prediction based on Monte Carlo simulation of the S-N curve scatter and Miner sum variability, enabling the TSII to incorporate calibrated uncertainty bounds that support risk-based maintenance decision making.

  TOWER-CORE: Structural Integrity Assessment and Safety Governance for Vertical Tower Systems · 2026 · DOI

• Technical Report · CC BY 4.0 · github.com/[username]/EVM-Insight 11 EVM-Insight — Technical Report · Zenodo Preprint · 2026 STRUCTURAL HEALTH MONITORING 11.1 Current Limitations (cid:127) Pre-experimental status: All performance claims are based on literature extrapolation and hardware specification analysis. No field measurements have been conducted. (cid:127) Ambient excitation dependency: EVM-Insight relies on ambient forcing to excite structural modes. Quiet environments or structures with high damping ratios may produce insufficient vibration amplitude. Artificial excitation (impact hammer) is not yet integrated. (cid:127) Single-frequency assumption: The current pipeline identifies one dominant frequency per ROI. Multi-modal vibration requires multiple band-pass filter passes with independent ROI selection. (cid:127) RPi 4 processing latency: EVM is post-processing, not real-time, on the RPi 4. Live EVM feedback during capture is a future extension. (cid:127) Homography registration limits: Very large baseline changes between viewpoints increase registration error. A shared reference marker protocol is under consideration. (cid:127) RPi 4 resource contention: EVM, sensor management, and WiFi hotspot run concurrently. Process priority management (nice, CPU affinity) will be implemented in Phase 6. 11.2 Future Work (cid:127) Phase-Based Motion Magnification (Wadhwa et al., 2013) as an alternative to intensity-based EVM. (cid:127) Multi-modal vibration analysis: simultaneous band-pass filtering at multiple frequency bands. (cid:127) POCO-side EVM acceleration via OpenCL/Vulkan compute shaders for ‡10 fps real-time feedback. (cid:127) Longitudinal monitoring: time-series archival of dominant frequencies across repeated sessions. (cid:127) Integration with public structural databases for automatic cross-reference with design specs. (cid:127) Federated multi-inspector deployment: multiple POCO devices sharing one RPi 4 infrastructure.

  EVM-Insight: A Low-Cost Eulerian Video Magnification System for Structural Health Monitoring · 2026 · DOI

• References Technical Report · CC BY 4.0 · github.com/[username]/EVM-Insight 2 EVM-Insight — Technical Report · Zenodo Preprint · 2026 STRUCTURAL HEALTH…

  EVM-Insight: A Low-Cost Eulerian Video Magnification System for Structural Health Monitoring · 2026 · DOI

• 1. Multi-site generalization. Extend deployments to multiple wharves and environmental regimes to test cross-asset transferability and robustness to distribution shift. 2. Compute-efficient twins. Develop reduced-order/ surrogate DTs with event-triggered updating and edge execution to retain low latency while cutting energy and cost. 3. Interpretable, uncertainty-aware ML. Integrate calibrated probabilities, conformal prediction, and DT-consistency checks to produce explanations and reliable risk bounds for operators and regulators. 4. Standardized benchmarks. Release harmonized protocols (EOV handling, chronological/LOP splits, Dev-only thresholding) and shared labeled datasets to enable reproducible, apples-to-apples comparisons across SHM frameworks. 5. Sensing durability and autonomy. Investigate energy harvesting, redundancy, and automated quality control (noise-floor/PSD drift/coherence tests) for long-term offshore operation. In summary, the proposed framework delivers high diagnostic accuracy (F1 0.95; AUC 0.98), rapid response (≈45 s latency), and tight DT–field agreement (MAC ≥ 0.92; frequency deviation within ±3%), offering a scalable path to next-generation, predictive condition monitoring for critical maritime infrastructure.

  Field-Validated Digital Twin–Enabled Structural Health Monitoring for Offshore High-Pile Wharves · 2026 · DOI

• The practical integration of component-level damage classification outputs into holistic building safety assessment lacks formalized decision-fusion methodologies; specifically, the field requires explicit frameworks for aggregating component-level damage grades (e.g., crack width quantification, column damage classification) into holistic safety tags while quantifying confidence levels and managing discrepancies between AI predictions and expert judgment in operational decision-support contexts.

  AI-Driven Post-Earthquake Rapid Structural Damage Assessment: A Scoping Review of Component-Level and Holistic Approaches · 2026 · DOI

• The operational deployment of AI-driven RSDA systems requires rigorous validation protocols on external test sets from entirely different seismic events beyond laboratory-controlled conditions; while some studies demonstrate this practice, systematic validation frameworks are needed to assess model robustness across unpredictable post-disaster field conditions, variable lighting, occlusions, and structural configurations that differ from training data distributions.

  AI-Driven Post-Earthquake Rapid Structural Damage Assessment: A Scoping Review of Component-Level and Holistic Approaches · 2026 · DOI

• The field must establish standardized, open-source benchmarks specifically for post-earthquake structural damage assessment beyond existing general crisis datasets; while initiatives like the Earthquake Image Dataset (EID) and EIDSeg provide pixel-level annotations for structural damage, comprehensive multi-event benchmarks with high-fidelity annotations covering diverse seismic events, building typologies, and environmental conditions are needed to enable reproducible and comparable component-level and holistic model performance evaluation.

  AI-Driven Post-Earthquake Rapid Structural Damage Assessment: A Scoping Review of Component-Level and Holistic Approaches · 2026 · DOI

• A fundamental disconnect exists between surface-level visual damage detection achieved by AI models and quantitative structural performance assessment; while studies successfully demonstrate crack, spalling, and collapse detection through instance segmentation and classification, these visual outputs rarely translate into actionable engineering metrics such as stiffness degradation, residual drift capacity, or yield strength for holistic building safety classification.

  AI-Driven Post-Earthquake Rapid Structural Damage Assessment: A Scoping Review of Component-Level and Holistic Approaches · 2026 · DOI

• Fine-grained damage detection remains an open challenge in component-level AI models for RSDA, specifically the reliable identification of hairline cracks, differentiation between multiple intersecting crack patterns, and distinction between minor and moderate damage levels. These subtle or geometrically complex cases limit the application of AI tools for early-stage damage detection where nascent damage indicators are critical for holistic structural assessment.

  AI-Driven Post-Earthquake Rapid Structural Damage Assessment: A Scoping Review of Component-Level and Holistic Approaches · 2026 · DOI

• Current AI-driven post-earthquake rapid structural damage assessment datasets suffer from severe class imbalance, with rare but critical damage conditions such as 'severe damage' and 'collapse' underrepresented compared to abundant 'undamaged' or 'slightly damaged' examples. This imbalance causes models to achieve high overall accuracy while failing to reliably identify catastrophic cases, where false negatives carry prohibitively high operational costs in real-world RSDA applications.

  AI-Driven Post-Earthquake Rapid Structural Damage Assessment: A Scoping Review of Component-Level and Holistic Approaches · 2026 · DOI

• When edges of the FE Model are anti-nodal points at the first mode when undamaged, some damage classes show zero F1-scores, indicating poor predictability for high severity damage at certain locations.

  Leveraging finite element model with transfer learning for data-driven-based structural damage identification · 2026 · DOI

• The effectiveness of SMIDA reduces as average deviation increases beyond 30%, suggesting the framework may not be effective when undamaged mode shapes have less heaving motion.

  Leveraging finite element model with transfer learning for data-driven-based structural damage identification · 2026 · DOI

• Further investigation on the optimum hyperparameters for SMIDA and the SMIDA-applicable range is needed to further validate the relationship between average deviation and identification accuracy.

  Leveraging finite element model with transfer learning for data-driven-based structural damage identification · 2026 · DOI

• Further investigation on the optimum parameters of SMIDA needs to be done to improve the unseen damage identification accuracy of Case 2, where training does not involve the transformed few-shot real data.

  Leveraging finite element model with transfer learning for data-driven-based structural damage identification · 2026 · DOI

• However, collecting vibration signals related to structural damage poses certain challenges, which can undermine the accuracy of the identification results produced by data-driven SDI methods in scenarios where data is scarce.

  Damage identification of steel bridge based on data augmentation and adaptive optimization neural network · 2024 · DOI

• The concept of PBSHM via TL is starting to be applied to bridge SHM, overcoming a significant limitation of traditional data- based SHM approaches, where inferences are limited to novelty detection in the absence of labelled training data.

  Transfer learning in bridge monitoring: Laboratory study on domain adaptation for population-based SHM of multispan continuous girder bridges · 2024 · DOI

• By developing a tailor-made vibration monitoring system for a hospital, this paper presents the monitoring and assessment of vibration impact on ultraprecision equipment throughout the whole construction period of a hospital expansion project, which, to the best of the authors’ knowledge, has not been reported in the literature.

  Monitoring and Assessment of Vibration Impact on Ultraprecision Equipment in a Hospital throughout a Whole Construction Period · 2023 · DOI

• The method is based on the assumption that the modified maximum strain value caused only by the axle loads may be easily used to identify the load of moving vehicles by eliminating the influence of these axle parameters from the peak value, which is not limited to a specific type of bridges and can be applied in conditions, where there are multiple moving vehicles on the bridge.

  Moving Load Identification with Long Gauge Fiber Optic Strain Sensing · 2021 · DOI

• The exact practical computation of modal damping is still an open issue, often leading to biased estimates since the errors are coming from every step in EFDD procedures and mainly due to signal processing.

  Enhanced frequency domain decomposition algorithm: a review of a recent development for unbiased damping ratio estimates · 2018 · DOI

• While deep learning is powerful for system identification, deterministic approaches lack reliable uncertainty quantification and can yield physically inconsistent results.

  Uncertainty-aware damage identification in short-span bridges via physics-informed variational autoencoder · 2026

• Conventional identification techniques often depend on subjective judgment and lack systematic uncertainty quantification, leading to inconsistent results and limited reproducibility, which ultimately hampers automation.

  Automatic modal parameter identification and associated uncertainty quantification for civil structures via SSI-COV and Deep Learning · 2026 · DOI

• Although the optimal design of TMDs has been investigated abundantly in the last few years, the effectiveness of TMDs in use has not been thoroughly studied.

  Effectiveness Assessment of TMDs in Bridges under Strong Winds Incorporating Machine-Learning Techniques · 2022 · DOI

• The results indicate that the presented methodology will enable engineers to use the updated structural model to determine the reserved capacity and remaining service life of structural elements, though further studies on methods to improve mesh generation and defect quantification are warranted.

  Damage Detection and Finite-Element Model Updating of Structural Components through Point Cloud Analysis · 2018 · DOI

Most-cited papers in Structural Health Monitoring Techniques

• Machine learning-based seismic response and performance assessment of reinforced concrete buildings · Archives of Civil and Mechanical Engineering · 2023 · 199 citations
• Optimization-based stacked machine-learning method for seismic probability and risk assessment of reinforced concrete shear walls · Expert Systems with Applications · 2024 · 146 citations
• Intelligent-inspired framework for fatigue reliability evaluation of offshore wind turbine support structures under hybrid uncertainty · Ocean Engineering · 2024 · 138 citations
• Missing measurement data recovery methods in structural health monitoring: The state, challenges and case study · Measurement · 2024 · 124 citations
• GNN-LSTM-based fusion model for structural dynamic responses prediction · Engineering Structures · 2024 · 106 citations
• Deep Neural Network and Evolved Optimization Algorithm for Damage Assessment in a Truss Bridge · Mathematics · 2024 · 105 citations
• INFRARED THERMOGRAPHY FOR DETECTING DEFECTS IN CONCRETE STRUCTURES · Journal of Civil Engineering and Management · 2018 · 105 citations
• Towards vibration-based damage detection of civil engineering structures: overview, challenges, and future prospects · International Journal of Mechanics and Materials in Design · 2024 · 98 citations
• A novel unsupervised deep learning approach for vibration-based damage diagnosis using a multi-head self-attention LSTM autoencoder · Measurement · 2024 · 86 citations
• An integrated deep neural network model combining 1D CNN and LSTM for structural health monitoring utilizing multisensor time-series data · Structural Health Monitoring · 2024 · 82 citations

Most recent work

• Dynamic monitoring of bridges subjected to freezing-induced variability by kernel-enhanced deep anomaly detection · Engineering Structures · 2026
• Leveraging finite element model with transfer learning for data-driven-based structural damage identification · Journal of Vibration Engineering & Technologies · 2026
• Deep learning-based realtime multiload response prediction and inverse analysis of offshore bridges · Engineering Structures · 2026
• TOWER-CORE: Structural Integrity Assessment and Safety Governance for Vertical Tower Systems · Zenodo (CERN European Organization for Nuclear Research) · 2026
• Understanding the spectral acceleration amplification factors of reinforced concrete frames through explainable artificial intelligence · Computers & Structures · 2026
• AI-Driven Post-Earthquake Rapid Structural Damage Assessment: A Scoping Review of Component-Level and Holistic Approaches · Archives of Computational Methods in Engineering · 2026
• Uncertainty law for modal identification using free vibration data · Mechanical Systems and Signal Processing · 2026
• Deep learning-based time series classification for damage assessment in multi-layer composite structures · Engineering Computations · 2026
• Physics-informed neural operator for forecasting vehicle-induced bridge vibration trajectories · Engineering Applications of Artificial Intelligence · 2026
• Optimizing Subpixel Vision-based Displacement Measurement for Structural Health Monitoring Applications · International Journal of Pattern Recognition and Artificial Intelligence · 2026

Related topics in Engineering