Narrow Results

The vehicle scanning method (VSM), originally known as the indirect method, is an efficient method for bridge health monitoring that utilizes mainly the responses collected by the moving test vehicles. This method offers the advantage of mobility, efficiency, and cost-effectiveness as it requires only one or a few vibration sensors mounted on the test vehicle, eliminating the need for deployment of numerous sensors on the bridge. Since its initial proposal by Yang and co-workers in 2004, the VSM has gained intensive attention from researchers worldwide. Over the past two decades, significant progress has been made in various aspects of the VSM, including the identification of bridge frequencies, mode shapes, damping ratios, damages, and surface roughness, as well as its application to railways. Previously, some review papers and the book Vehicle Scanning Method for Bridges were published on the subject. However, research on the subject continues to boom at a speed that cannot be adequately by existing review papers or book, as judged by the fast-increasing number of relevant publications. In order to provide researchers with an overall understanding of the up-to-date researches on the VSM, a state-of-the-art review of the related research conducted worldwide is compiled in this paper. Comments and recommendations will be provided at appropriate points, and concluding remarks, including future research directions, will be presented at the end of the paper.

Structural damage detection is crucial for ensuring the safety of civil building structures in operational environments. Recently, deep learning-based methods have gained increasing attention from engineers and researchers. The performance of conventional deep learning methods for structural damage detection relies on a large number of labeled training datasets. However, it is difficult or/and impossible to obtain sufficient datasets to cover various damage scenarios for in-service structures. A little research has been conducted to identify both the damage severity and location with limited labeled measurement data. A novel transfer learning-based method for structural damage identification with limited measurements has been proposed utilizing frequency response functions (FRFs) as the input. The real structure is regarded as the target domain and its numerical model is as the source domain. The samples for various damage scenarios are generated using the numerical model, and a designed deep convolutional neural network (CNN) is pre-trained. The knowledge of the pre-trained network is transferred to identify the damage location and severity of the real structure using limited measurement data. Numerical and experimental studies have been conducted on a three-story building structure to verify the performance of the proposed method. To understand transfer learning and model interpretability, the t-SNE feature visualization is adopted to show the feature distribution changes during transfer learning. Numerical and experimental results show that the proposed approach outperforms conventional CNN models, and it is effective and accurate in identifying structural damage location and severity in real structures with limited measurement data.

For civil engineering structures, structural damage usually occurs at limited positions in the preliminary stage of the structural failure. Compared with the numerous elements of the entire structure, the damaged elements are sparsely distributed in space. Based on this important prior information, this paper proposed to utilize a sparse Bayesian learning method to identify the damage to structures while considering the measurement noise and modeling error. The particle Swarm Optimization (PSO) algorithm was first introduced to address the associated computational efficiency issue, and the optimization performances of PSO and Sequential Quadratic Programming (SQP) algorithm in the process of model updating were compared, positive outcomes revealed that the PSO algorithm has the stronger searching ability and better robustness. To investigate the effectiveness and practicality of the sparse Bayesian learning with a PSO algorithm in structural damage detection, an asymmetrical frame in different scenarios (e.g. with single and multiple damages) was constructed in the laboratory. The encouraging results of the experimental case studies compellingly demonstrate that the presented methodology not only can detect the location and extent of structural damage with high precision and efficiency, but also can proficiently assess the posterior uncertainties associated with the damage detection results.

Recent advancements in structural health monitoring have been significantly driven by the integration of artificial intelligence technologies. This study employs a combination of supervised machine learning techniques, including classification and regression, to accurately detect and localize local thickness reduction defects in a cantilever beams. Our approach utilizes a dataset of 100 signals, comprising 84 defective and 16 healthy states of the beam’s free side displacement, for training machine learning models. Signal processing involves the application of five distinct mode decomposition methods to decompose each signal into its Intrinsic Mode Functions (IMFs). Additionally, four dimensionality reduction methods have been used to reduce the dimensions of the signals. Feature extraction is performed using seven frequency domain, two time domain, and three time–frequency domain methods to capture pertinent patterns and characteristics within the signals. We evaluate the performance of five classification methods and 10 regression methods to predict the location of defects. Our results demonstrate the efficacy of combining specific feature extraction and dimensionality reduction techniques with classification methods, achieving multi-class classification accuracies of up to 99.55%. Moreover, regression methods, particularly the Bayesian ridge regressor, exhibit high accuracy in predicting defect locations, with an $R^2$ value of 99.94% and minimal Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) values. This study highlights the potential of integrating regression and classification-based machine learning approaches for precise damage detection and localization in beam structures.

Starting from the derivation of an analytical method to expand measured static displacement data to full degrees of freedom, this study proposes a damage detection method to detect the damage of damaged beam by introducing displacement curvature and damage factor (DF). The validity of the proposed method is evaluated in damaged beam system that two simple beams are perpendicularly interconnected at a point.

Traditional machine learning requires users to have a strong ability to control features and distance calculation formulas, especially in the use of support vector machine SVM and nearest neighbor KNN. Traditional machine learning uses PCA in feature extraction will actually lead to Information is lost. In order to solve the problem of low optical film damage detection rate of traditional methods, a new method is proposed in this paper based on a convolutional neural network instead of traditional machine learning to classify CCD images with different damage degrees of SiO₂ film and K9 glass. First, film images are collected by online CCD, and the proposed algorithm is designed to extract the image characteristic parameters of the film microscopic images, filter denoising, and run binarization to analyze film images. Second, gray values of images are extracted and classified by unsupervised learning. Finally, the film microscopic images under the microscope are analyzed. The experimental results show that the defect positions on the images can be detected after the images are detected and processed by a convolution neural network, binarization, and connected domains. The defective parts can be intercepted from the images, and the data related is saved for damage type determination. The average classification rate is over 99%, which is better than the traditional method by 9.1%. Therefore, it has a high application value.

Based on the wave theory and propagation characteristics of Lamb wave in the thin plate, Lamb wave is excited by signal actuator in the thin plate and signal sensor is received in real time. Employing the superior features of dispersion curve produced the Lamb wave and combining the dichotomy iterative principle with MATLAB tools to solve Lamb wave dispersion equation, the propagation characteristics of the Lamb wave in the thin plate are plotted. It is verified that the excited Lamb wave has multimode characteristics in a certain thickness, and the different modes are related to the frequency-thickness. The wavelength-thickness ratio is defined to simplify the calculation and provide convenience for the frequency bandwidth interception of Lamb wave. The results of the theoretical analysis show that the propagation multimode of Lamb and the distribution characteristics of dispersion curve can effectively simulate the transmission information of damage signal-excited thin plate, and identify damage region and determine the damage degree by $S_0$ and $A_0$ mode. The experimental results indicate that the established method to determine damage degree of thin plate attached PZT transducers based on Lamb wave active monitoring technology can achieve a certain precision, and can make a rough quantitative recognition of damage position.

This paper presents an effective way in damage detection of beam structures using the wavelet analysis along with the general beam solution. Two case studies are considered: (1) a clamped beam with a damage point of zero bending moment; and (2) a simply supported beam with a transverse open crack. The proposed method is capable of revealing the precise damage locations which is generally difficult to be identified using the standard eigenvalue analysis.

A statistical method using frequencies of structures under control is proposed for detecting damage. In the study, feedback control based on independent modal space control is first used to assign the pole of the system under detection intentionally. Then the prescribed characteristic frequencies of the structure under control, which may be more sensitive to damage, are obtained and further employed to constitute a sensitivity-enhanced damage indicator (SEDI). The principle of sensitivity-enhancing feedback control for damage detection of multi-degree-of-freedom systems is elaborated. To overcome the effect of measurement noise on modal frequencies, a hypothesis test involving the t-test that utilizes the SEDI is employed to estimate the occurrence of damage, while a statistical pattern recognition method that uses the feature vectors including the SEDI is employed to locate damage. Based on the perturbation theory, the feature vectors are normalized in order to eliminate the effect of damage extent on damage localization. The proposed method is verified by examples including a three-span continuous beam with a single damaged element and the IASC-ASCE benchmark structure with a single damaged brace. Simulation results show that, by using the frequencies of the structures under control, the proposed damage indicators are more sensitive to damage and are capable of detecting and locating small damage of structures.

Presented herein is an experiment that aims to investigate the applicability of the wavelet transform to damage detection of a beam–spring structure. By burning out the string that is connected to the cantilever beam, high-frequency oscillations are excited in the beam–spring system, and there results an abrupt change or impulse in the discrete-wavelet-transformed signal. In this way, the discrete wavelet transform can be used to recognize the damage at the moment it occurs. In the second stage of damage detection, the shift of frequencies and damping ratios is identified by the continuous wavelet transform so as to ensure that the abrupt change or impulse in the signal from the discrete wavelet transform is a result of the damage and not the noise. For the random forced vibration, the random decrement technique is used on the original signal to obtain the free decaying responses, and then the continuous wavelet transform is applied to identify the system parameters. Some developed p version elements are used for the parametric studies on the first stage of health monitoring and to find the damage location. The results show that the two-stage method is successful in damage detection. Since the method is simple and computationally efficient, it is a good candidate for on-line health monitoring and damage detection of structures.

Studied herein are the signatures of nonlinear vibration characteristics of damaged reinforced concrete structures using the wavelet transform (WT). A two-span RC slab built in 2003 was tested to failure in the laboratory. Vibration measurements were carried out at various stages of structural damage. The vibration frequencies, mode shapes, and damping ratios at each loading stage were extracted and analyzed. It is found that the vibration frequencies are not sensitive to small damages, but are good indicators when damage is severe. The dynamic responses are also analyzed in the time–frequency domain by WT and the skeleton curve is constructed to describe the nonlinear characteristics in the reinforced concrete structures. The results show that the skeleton curves are good indicators of damage in the reinforced concrete structures because they are more sensitive to small damages than vibration frequencies.

The reliability of a structural health-monitoring system is very dependent on the quality of signals that are acquired and fed into the damage detection algorithm. Developed herein is an algorithm for sensor validation in the context of the damage locating vector (DLV) method for detecting damage in structures. Given the signals from ns sensors, ns combinatorial sets of signals from (ns - 1) sensors are formed. For each set, the change in the flexibility matrix relative to that of the reference or undamaged structure is computed and singular value decomposition is performed to estimate the number of nonzero singular values (NZV). The set that produces the smallest NZV is associated with healthy sensors whereas sensors that do not belong to that combination are suspected to be faulty. The performance of the proposed algorithm is illustrated using both simulated and experimental data obtained from a 3D modular truss structure monitored by sensors, some of which are faulty.

This paper addresses a proficient strategy for detection of structural damages in details using the variations of eigenvalues and eigenvectors. There are two concerns in this study. First, the severity of damage can vary within the damaged elements; second, it is possible that the damage extents do not exactly match the pre-generated finite element mesh. The first concern forms the motivation for employing the proper damage functions to model the elemental damages, and the second for considering the nodal positions as design variables. To obtain the design variables, an improved genetic algorithm is introduced in which two new operators are embedded. This strategy is applied to a beam and a plate structure as the cases of study. The results demonstrate the applicability and efficiency of the proposed algorithm in elaborate damage detections.

Shear connectors are generally used to link the slab and girder together in slab-on-girder bridge structures. Damage of shear connectors in such structures will result in shear slippage between the slab and girder, which significantly reduces the load-carrying capacity of bridges. A damage detection approach based on transmissibility in frequency domain is proposed in this paper to identify the damage of shear connectors in slab-on-girder bridge structures with or without reference data from the undamaged structure. The transmissibility, which is an inherent system characteristic, indicates the relationship between two sets of response vectors in frequency domain. Measured input force and acceleration responses from hammer tests are analyzed to obtain the frequency response functions at the slab and girder sensor locations by the experimental modal analysis. The transmissibility matrix that relates the slab response to the girder response is then derived. By comparing the transmissibility vectors in undamaged and damaged states, the damage level of shear connectors can be identified. When the measurement data from the undamaged structure are not available, a study with only the measured response data in the damaged state for the condition assessment of shear connectors is also conducted. Numerical and experimental studies on damage detection of shear connectors linking a concrete slab to two steel girders are conducted to validate the accuracy and efficiency of the proposed approach. The results demonstrate that the proposed method can be used to identify shear connector damages accurately and efficiently. The proposed method is also applied to the condition evaluation of shear connectors in a real composite bridge with in-field testing data.

Structural damage detection and vibration control have been actively studied in the past decades. However, in most previous investigations, these two aspects have been treated separately according to their individual primary objectives pursued. Although structural identification and model updating have been used in the adaptive control techniques, it is still necessary but challenging to study the on-line integration of structural damage detection and optimal vibration control. In this paper, a technique is proposed for such purpose. First, on-line structural parameter identification and on-line detection of the onset, locations and extents of structural damage of controlled structures with partial measurements of structural acceleration responses is studied. An algorithm is proposed based on the extended Kalman estimator/Kalman estimator and error tracking. Then, the updated structural models are on-line integrated with the instantaneous optimal control scheme to reach the goal of optimal active structural control of the undamaged/damaged structures. Numerical simulation results with different structural damage patterns illustrate the performances of the proposed technique for the on-line integration of structural damage detection and active optimal vibration control.

Traditional structural system identification and damage detection methods use vibration responses under single excitation. This paper presents an auto/cross-correlation function-based method using acceleration responses under multiple ambient white noise or impact excitations. The auto/cross-correlation functions are divided into two parts. One is associated with the structural parameters and the other with the energy of the excitation. These two parts are updated sequentially using a two-stage method. Numerical and experimental studies are conducted to demonstrate the accuracy and robustness of the proposed method. The effects of measurement noise and number of measurement points on the identification results are also studied.

This paper presents a genetic algorithm (GA)-based method to identify the damage of girder bridges from the response of a vehicle moving over the bridge. The continuous wavelet transform-based method works when the surface is smooth but the identification becomes difficult when the road surface is rough. To deal with this problem, the identification process is formulated as an optimization problem and a guided GA is used to search for the global optimal value. The vertical accelerations of the vehicle running over the bridge at the intact and damaged states are used to identify the occurrence and location of the damage. Frequencies of the bridge at the intact and damaged states can be extracted from these responses, from which the frequency-based method can roughly estimate the possible locations of the damage. These locations are not unique as frequencies alone are insufficient to identify the damage location. However these initial results can be used to narrow down the search region on which the GA can focus. Numerical study shows that the strategy can identify the damage location for simply supported and continuous girder bridges even though road surface roughness and measurement noise are taken into account.

This paper presents a novel approach for structural damage detection and estimation using incomplete noisy modal data and artificial neural network (ANN). A feed-forward back propagation network is proposed for estimating the structural damage location and severity. Incomplete modal data is used in the dynamic analysis of damaged structures by the condensed finite element model and as input parameters to the neural network for damage identification. In all cases, the first two natural modes were used for the training process. The present method is applied to three examples consisting of a simply supported beam, three-story plane frame, and spring-mass system. Also, the effect of the discrepancy in mass and stiffness between the finite element model and the actual tested dynamic system has been investigated. The results demonstrated the accuracy and efficiency of the proposed method using incomplete modal data, which may be noisy or noise-free.

The identification of railway ballast damage under a concrete sleeper is investigated by following the Bayesian approach. The use of a discrete modeling method to capture the distribution of ballast stiffness under the sleeper introduces artificial stiffness discontinuities between different ballast regions. This increases the effects of modeling errors and reduces the accuracy of the ballast damage detection results. In this paper, a continuous modeling method was developed to overcome this difficulty. The uncertainties induced by modeling error and measurement noise are the major difficulties of vibration-based damage detection methods. In the proposed methodology, Bayesian probabilistic approach is adopted to explicitly address the uncertainties associated with the identified model parameters. In the model updating process, the stiffness of the ballast foundation is assumed to be continuous along the sleeper by using a polynomial of order N. One of the contributions of this paper is to calculate the order N conditional on a given set of measurement utilizing the Bayesian model class selection method. The proposed ballast damage detection methodology was verified with vibration data obtained from a segment of full-scale ballasted track under laboratory conditions, and the experimental verification results are very encouraging showing that it is possible to use the Bayesian approach along with the newly developed continuous modeling method for the purpose of ballast damage detection.

This paper presents an innovative technique for structural damage detection based on time series analysis with feedback controllers incorporated into the structure. The sensitivity of autoregressive (AR) coefficients to element stiffness is first derived, and it is proposed that the sensitivity of the AR coefficients can be enhanced by intentionally assigning the poles of the detection system. Finally, an $\bar{x}$ control chart is constructed based on the sensitivity-enhanced AR coefficient. Identification of the occurrence of damage is achieved by monitoring statistically significant changes in the control chart. The proposed methodology is validated by examples including a cantilever Euler beam and Phase I of the IASC-ASCE benchmark structure. The simulation results show that by utilizing the sensitivity-enhanced AR coefficients, the control charts are more sensitive to damage and are capable of detecting small levels of structural damage even in the presence of measurement noise.