Drive-by damage detection and localization exploiting continuous wavelet transform and multiple sparse autoencoders

Structural ageing and material deterioration require infrastructure managers to continuously seek for improved solutions for bridge condition management. In the last two decades, vehicle-assisted bridge monitoring has emerged among researchers and engineers as a promising tool to support visual inspections, being a cost-efficient alternative to direct Structural Health Monitoring systems. In this work, the authors present a sparse-autoencoder-based damage detection methodology which exploits the vertical acceleration of train’s leading bogie to assess bridge health condition. The bridge under analysis in this work is a Warren truss bridge, whose FE model was designed based on the technical drawings of an actual structure, which belongs to the Italian railway line, and then validated through dynamic measurements. Raw bogie vertical accelerations are preprocessed through Continuous Wavelet Transform, allowing for the extraction of a specific frequency region of interest, governed by the modular configuration of the bridge as well as the forward speed of the convoy. Starting from the average curve of the computed wavelet coefficients, bridge health status is assessed through the use of a sparse autoencoder exploiting multiple train transits and two different damage indices. First of all the Hotelling’s statistic computed at the latent space level, and, secondly, the batch mean absolute reconstruction error. Different damage scenarios and intensities are tested in this work, considering the partial failure of stringers or cross-girder members due to corrosion, modelled as material mass loss and stiffness reduction. A large set of simulations allowed for testing the robustness of the methodology against operational variables, such as travelling speed, geometrical track irregularity evolution, and different weights of the convoy. Globally, promising damage detection performances were obtained when considering batches of 40 trains, even in presence of measurement noise and speed estimation inaccuracies.

Fukunaga-Koontz feature transformation for statistical structural damage detection and hierarchical neuro-fuzzy damage localisation

Recently, the authors proposed computationally attractive algorithms to determine the location and extent of structural damage for undamped structures assuming damage results in a localized change in stiffness properties. The algorithms make use of a finite-element model and a subset of measured eigenvalues and eigenvectors. The developed theories approach the damage location and extent problem in a decoupled fashion. First, a theory is developed to determine the location of structural damage. With location determined, a damage extent theory is then developed. The damage extent algorithm is a minimum rank perturbation, which is consistent with the effects of many classes of structural damage on a finite-element model. In this work, the concept of the minimum rank perturbation theory (MRPT) is adopted to determine the damage extent on the mass properties of an undamped structure. In addition, the MRPT is extended to the case of proportionall y damped structures. For proportionally damped structures, the MRPT is used to find the damage extent in any two of the three structural property matrices (mass, damping, or stiffness). Finally, illustrative case studies using both numerical and actual experimental data are presented. HE advent of the Space Shuttle has prompted considerable attention to the design and control of large space structures. Due to the large size and complexity of envisioned structures, as well as the use of advanced materials to reduce structural weight, it may become necessary to develop a structural health monitoring system to detect and locate structural damage as it occurs. From experience gained in the machinery health monitoring field, one would expect the vibration signature of the structure, either frequency response functions and/or modal parameters, to provide useful information in determining the location and extent of structural damage. Assume that a refined finite element model (FEM) of the structure has been developed before damage has occurred. By refined, we mean that the measured and analytical modal properties are in agreement. Next, assume that at a later date some form of structural damage has occurred. If significant, the damage will result in a change in the structures modal parameters. The question is: can the discrepancy between the original FEM modal properties and postdamage modal properties be used to ascertain structural damage? Most prior work in damage detection has used the general framework of FEM refinement (system identification) in the development of damage assessment algorithms. The motivation behind the development of FEM refinement techniques is based on the need to validate engineering FEMs before their acceptance as the basis for final design analysis. The standard problem has been to seek a refined FEM that is as close to the original FEM and whose modal properties are in agreement with those that are measured subject to various constraints such as symmetry and sparsity preservation. A considerable amount of work in this area has been

Structural health monitoring (SHM) is essential for ensuring the durability and safety of steel bridges, which are considered as critical part of modern infrastructure. Ensuring the integrity of bridges is vital for maintaining their functionality and safety. Considerable research has been conducted on predicting damage location in multi-span steel bridges, while relatively limited attention has been given to the identification of damage in single-span steel bridges, and there exists a notable gap in the literature regarding the quantification of damage based on severity. This study presents a novel approach combining Modal Strain Energy-based damage indices with extensively trained data-driven Artificial Neural Networks (ANNs) to achieve highly accurate damage location and quantify this damage in terms of severity in single-span steel bridges. The research integrates a finite element (FE) model as the case study of the Alamosa Canyon Bridge in Sierra County, New Mexico. The validated FE model was used to compute damage indices based on the first two bending mode shapes. The damage index (DI) of each mode for different damage scenarios simulated on the FE model is then combined using the absolute value method to pinpoint damage locations. ANNs are employed to predict severity based on these damage indices, leveraging a comprehensive dataset generated through cubic spline interpolations and FE simulations. The DI was used as an input parameter for the ANNs to estimate damage. This methodology highlights the reduced computational effort using a streamlined method for structural health monitoring that maintains high accuracy in severity prediction and damage detection. This study enhances the infrastructure safety and maintenance strategies by offering a practical framework for steel bridges.

Transmissibility of non-linear output frequency response functions with application in detection and location of damage in MDOF structural systems

This paper presents an enhanced version and the validation of a recently proposed methodology for earthquake-induced damage detection and localization in masonry towers by using long-term vibration monitoring data. The proposed enhanced method is based on continuous operational modal analysis through stochastic subspace identification, dynamic multiple linear regressive analysis to remove the effects of changing environmental conditions and finite element (FE) model updating. Any anomalous frequency deviation from normal conditions resulting from an earthquake triggers the damage localization process. This task is performed by solving an inverse FE model calibration problem, where equivalent elastic properties of macrostructural elements are identified by minimizing an objective function considering experimentally identified and numerically predicted damage-induced decays in natural frequencies and changes in eigenvector components. To minimize the computational effort of this calibration procedure, a quadratic surrogate model is constructed using a tuned numerical FE model of the structure. The methodology is validated through application to the “Sciri Tower”, an historic civic masonry tower located in Perugia, Italy, that has been continuously monitored by the authors for more than 1 year. The validation is carried out by using simulated damage scenarios and a set of real far-field earthquake data and is based on a FE model of the tower including surrounding buildings, calibrated on the basis of the measured data from ambient vibration tests. The results demonstrate that the proposed procedure is capable of correctly detecting and localizing earthquake-induced damages.

Damage detection and localization in civil or mechanical structures is a subject of active development and research. A few vibration‐based methods have been developed so far, requiring, for example, modal parameter estimates in the reference and damaged states of the investigated structure, and for localization in addition a finite element model. For structures in operation, temperature has been shown to be a major nuisance to the efficiency of such methods because the modal parameters are varying not only with damage but also due to temperature variations. For detection, a few rejection approaches have been developed. Besides the increased complexity, environmental variation is hardly taken into account in localization approaches. In this paper, we propose a sensitivity‐based correction of the identified modal parameters in the damaged state with respect to the temperature field in the reference state, on the basis of a sensitivity analysis with respect to temperature dependent Parameters of the finite element model in the reference state. The approach is then applied to the stochastic dynamic damage locating vector method, where its improved performance under nonuniform temperature variations is shown in a numerical application on a beam.

Accurate structural damage localization in engineering is still a challenge due to high noise and low accuracy of the structural finite element model. With the most basic type of construction, the simply supported beam is the most widely used type of structures in short and medium span bridges. Therefore, it is imperative to study their damage detection methods, which can be used in engineering. To solve this problem, an alternative method for damage detection is proposed in this study on the basis of fractal theory and curvature mode shapes method: the curvature difference probability method of waveform fractal dimension. The influence of different amplitudes of excitation on results and the robustness against noise are also discussed. Based on a simply supported beam model in the laboratory, both the experimental and numerical results using the technique under step or pulse excitation indicate that the proposed method can be used in damage localization very well, and it exhibits high-noise insusceptibility: it is still feasible even if the noise level is up to 15%, which lays a good foundation for engineering application. Moreover, the proposed technology needs no finite element model of the detected structures. Therefore, it is quite simple in the procedure.

Damage detection for beam-like structures using the normalized curvature of a uniform load surface

Impact of time synchronization error on the mode-shape identification and damage detection/localization in WSNs for structural health monitoring

Damage detection is a challenging, complex, and at the same time very important research topic in civil engineering. Identifying the location and severity of damage in a structure, as well as the global effects of local damage on the performance of the structure are fundamental elements of damage detection algorithms. Local damage detection is essential for structural health monitoring since local damages can propagate and become detrimental to the functionality of the entire structure. Existing studies present several methods which utilize sensor data, and track global changes in the structure. The challenging issue for these methods is to be sensitive enough in identifYing local damage. Autoregressive models with exogenous terms (ARX) are a popular class of modeling approaches which are the basis for a large group of local damage detection algorithms. This study presents an algorithm, called Influence-based Damage Detection Algorithm (IDDA), which is developed for identification of local damage based on regression of the vibration responses. The formulation of the algorithm and the post-processing statistical framework is presented and its performance is validated through implementation on an experimental beam-column connection which is instrumented by dense-clustered wired and wireless sensor networks. While implementing the algorithm, two different sensor networks with different sensing qualities are utilized and the results are compared. Based on the comparison of the results, the effect of sensor noise on the performance of the proposed algorithm is observed and discussed in this paper.

This study presents a modal-based damage detection technique for framed structures. A framed structure is tested using a vibration-based method and the mode shapes corresponding to undamaged and damaged states of the structure are obtained. The damage to the framed structure is induced by reducing the stiffness of columns at a particular storey and tested using the shaking table. The corresponding finite element (FE) model was verified for its natural frequency with the experimental model. Moreover, a series of studies were performed by concentrating the damage to a smaller scale by including spot damage on the framed structure and varying the damage location and intensity. The model was analysed using finite element software for these tests and mode shape slope differences and mode shape curvature differences were generated. Derivatives of mode shape slopes were plotted and correlated with the damage location. Mode shape slope is discontinuous and becomes a maximum value close to the damaged location. Mode shape curvatures also show a maximum difference near the damaged location. Finite element model, as well as the experimental investigation, shows the competence of the proposed approach in structural damage identification.

Damage Detection of Bridge Structures Using Modal Flexibility under Temperature Variations

Bridge damage detection using precise vision-based displacement influence lines and weigh-in-motion devices: Experimental validation

A robust technique for damage identification of marine fiberglass rectangular composite plates using 2-D discrete wavelet transform and radial basis function networks