Features For Damage Detection Research Articles

Robustness of wavelet features for damage detection with neural network and digital twin

Recent studies demonstrate that the wavelet energy–based features are highly sensitive to local damages, and accurate damage identification could be achieved by integrating such features with neural networks. However, the robustness of the features in practical applications and the feasibility of generating damage information with a digital twin to facilitate training of neural networks have not been systematically investigated. This paper aims to provide new insight into the practicality of using wavelet energy–based features for accurate damage detection of structures. A systematic investigation into the tolerance of the normalised wavelet packet node energy features with neural network (NWPNE-NN) approach against measurement noises, excitations uncertainties and limited frequency range in the measured responses is carried out, with consideration of data fusion from multiple measurement points. The feasibility of a digital twin to simulate damage scenarios and generate training data for neural networks is explored, and the use of relative WPNE as a new feature is proposed. The results show that the NWPNE-NN approach is capable of detecting and quantifying the main damage in a structure. The neural networks trained by data generated from a well-calibrated digital twin is capable of identifying damage in the actual structure.

A Loewner-Based System Identification and Structural Health Monitoring Approach for Mechanical Systems

Data-driven structural health monitoring (SHM) requires precise estimates of the target system behaviour. In this sense, SHM by means of modal parameters is strictly linked to system identification (SI). However, existing frequency-domain SI techniques have several theoretical and practical drawbacks. This paper proposes using an input-output system identification technique based on rational interpolation, known as the Loewner framework (LF), to estimate the modal properties of mechanical systems. Pioneeringly, the Loewner framework mode shapes and natural frequencies estimated by LF are then applied as damage-sensitive features for damage detection. To assess its capability, the Loewner framework is validated on both numerical and experimental datasets and compared to established system identification techniques. Promising results are achieved in terms of accuracy and reliability.

Investigations of Ratio-Based Integrated Influence Lines as Features for Bridge-Damage Detection

Prestressed concrete bridges built between 1960 and 1990 no longer meet today’s requirements due to loads and increasing mileage of higher loads that have increased since the bridges were designed. Prestressed concrete bridges are representative of Germany’s existing bridges. In order to deal with the large number of ageing bridges, recalculations and measurements for control as well as bridge monitoring are an important means of support. For both, it is important to find features that are damage-sensitive as well as robust against measurement noise, vehicle parameters (dynamics, geometry, weight, etc.) and environmental influences (temperature, wind, etc.). In this paper, we present features for damage detection based on the influence line, which are investigated with respect to the above requirements by using the analytical solution of the Euler–Bernoulli beam and more complex numerical bridge simulations. In this context, we restrict ourselves to the damage caused by bending stress. The features are calculated on the basis of single vehicle crossings over the bridge for the strain in the longitudinal direction as well as for the deflection of the bridge at different sensor positions. The ratio-based features are compared with raw data and natural frequencies in a classification. Additionally, the sensor positioning is considered. The investigations shows that the ratio-based integrated influence lines are equivalent to or better than the modal parameters, especially when noise and temperature changes are taken into account.

Unsupervised Learning Approach for Detection and Localization of Structural Damage using Output-only Measurements

Interrogation of the vibration data collected from the sensors embedded throughout the structure without relying on a finite element model of the system for monitoring the health of structural systems has received significant attention in the recent years especially with the current advancements in sensor technology. The data-driven methods explored within this context falls into the realm of statistical pattern recognition field requiring extraction of damage detection features and a statistical decision-making process for identification of damage. Machine learning algorithms provide statistical means for making such decisions. In this study, an unsupervised machine learning approach, one-class support vector machine (OC-SVM), requiring training data only from the undamaged state of the structure is explored for damage detection purposes. The coefficients of the autoregressive (AR) model are extracted as damage sensitive features and used as the required training data. The trained classifier is then used with the data obtained from the same structure at different damage states for classification. Damage detection in the form of recognizing outliers or anomalies not belonging to the target class, is followed by damage localization within the given sensor resolution using statistical means. Numerical simulations are performed on a truss and a beam structure with several damage scenarios illustrating the capabilities and the limitations of the proposed approach.

Impact of Groundwater Level Change on Natural Frequencies of RC Buildings

Structural health monitoring (SHM) provides an opportunity to assess and predict changes in the technical condition of structures during the operation of a building. Structural damage, as well as several operational and environmental conditions, causes changes in modal parameters. Temperature is the most popular environmental condition which is used for research. However, to the authors’ knowledge, this is the first investigation that highlights the effect of groundwater level change on the natural frequencies of the buildings and the impact of possible damage detection features. Groundwater level change can influence structural health monitoring measurements and cause faulty structural damage identification using vibration-based methods. This paper aims to analyse the impact of the groundwater level changes on the modal parameters of mid-rise reinforced concrete buildings. The modal parameters of mid-rise reinforced concrete buildings are determined using finite element (FE) models. Three different FE models of structural system types of nine-storey reinforced concrete (RC) buildings with shallow foundations are used to determine the impact of groundwater level fluctuation on the values of the buildings’ natural frequencies. Changes in the groundwater level have an impact on the natural frequencies of the mid-rise reinforced concrete buildings. This research proposes a new environmental condition that has to be considered to identify the structural damage using the vibration-based method. It is found that groundwater level rise causes a decrease in the natural frequency value. In this research, it is established that the influence of the groundwater level on the natural frequencies of the buildings can change abruptly, and there is a non-linear correlation between groundwater level change and natural frequencies of the buildings. The natural frequencies of the buildings can change under varying environmental conditions as well as in the case of structural damage. To identify structural damage in the long-term structural health monitoring measurements, it is recommended to select features which are sensitive to structural damage but are not affected by groundwater level change. Data normalisation and elimination using linear correlation methods can be used for short-term SHM under varying seasonal groundwater level change.

Reconstructing Element-by-Element Dissipated Hysteretic Energy in Instrumented Buildings: Application to the Van Nuys Hotel Testbed

AbstractThe authors propose a seismic monitoring framework for instrumented buildings that employs dissipated energy as a feature for damage detection and localization. The proposed framework emplo...

Early damage assessment in large-scale structures by innovative statistical pattern recognition methods based on time series modeling and novelty detection

Time series analysis and novelty detection are effective and promising methods for data-driven structural health monitoring (SHM) based on the statistical pattern recognition paradigm. However, processing substantially large volumes of vibration measurements may represent a serious limitation, especially for long-term SHM programs of large-scale civil structures. Moreover, shortcomings like the choice of an appropriate time series model in an automatic manner, the determination of optimal orders of the identified model and the classification of random high-dimensional features for damage detection, can strongly affect the performance of these approaches. This study is intended to propose statistical pattern recognition methods regarding time series modeling for feature extraction and novelty detection in feature classification in the presence of big data. These methods include an automatic model identification algorithm, an improved order determination approach and a hybrid distance-based novelty detection through a combination of Partition-based Kullback-Leibler divergence and Mahalanobis-squared distance. Experimental datasets relevant to a cable-stayed bridge are considered to validate the effectiveness of the proposed methods. Results demonstrate that: the AutoRegressive-AutoRegressive with eXogenous input (AR-ARX) model turns out to be the most suitable representation for feature extraction; the orders of this model are efficiently and automatically determined; the proposed novelty detection approach is highly successful in detecting damage, even in case of large volumes of random high-dimensional features.

Detection and Quantification of Damage in Metallic Structures by Laser-Generated Ultrasonics

The appearance of damage on metallic structures is inevitable due to complex working environments. Non-destructive testing (NDT) of these structures is critical to the safe operation of the equipment. This paper presents a non-destructive damage detection, visualization, and quantification technique based on laser-generated ultrasonics. The undamaged and damaged metallic structures are irradiated with laser pulses to produce broadband input ultrasonic waves. Damage to the structures plays the role of a nonlinear radiation source of new frequencies. Usually these new frequencies are too weak to be detected directly. Here, the state space predictive model is proposed to address the problem. Based on the recorded responses in the time domain, the state space attractors are reconstructed. Damage to the structures is shown to change the properties of the attractors. A nonlinear damage detection feature called normalized nonlinear prediction error (NNPE) is extracted from the state space to identify the changes in the attractors—and hence the damage. Furthermore, the damage is visualized and quantified using the NNPE values extracted from the entire area by using a laser scanning technique. Experimental results validate that the proposed technique is capable of detecting, visualizing and quantifying artificial damage to aluminum alloy plates and actual fatigue cracks to a twin-screw compressor body.

Ensembles of novelty detection classifiers for structural health monitoring using guided waves

Guided wave structural health monitoring uses sparse sensor networks embedded in sophisticated structures for defect detection and characterization. The biggest challenge of those sensor networks is developing robust techniques for reliable damage detection under changing environmental and operating conditions (EOC). To address this challenge, we develop a novelty classifier for damage detection based on one class support vector machines. We identify appropriate features for damage detection and introduce a feature aggregation method which quadratically increases the number of available training observations. We adopt a two-level voting scheme by using an ensemble of classifiers and predictions. Each classifier is trained on a different segment of the guided wave signal, and each classifier makes an ensemble of predictions based on a single observation. Using this approach, the classifier can be trained using a small number of baseline signals. We study the performance using Monte-Carlo simulations of an analytical model and data from impact damage experiments on a glass fiber composite plate. We also demonstrate the classifier performance using two types of baseline signals: fixed and rolling baseline training set. The former requires prior knowledge of baseline signals from all EOC, while the latter does not and leverages the fact that EOC vary slowly over time and can be modeled as a Gaussian process.

Algorithm for damage detection in wind turbine blades using a hybrid dense sensor network with feature level data fusion

Damage detection in wind turbine blades requires the ability to distinguish local faults over a global area. The implementation of dense sensor networks provides a solution to this local-global monitoring challenge. Here the authors propose a hybrid dense sensor network consisting of capacitive-based thin-film sensors for monitoring the additive strain over large areas and fiber Bragg grating sensors for enforcing boundary conditions. This hybrid dense sensor network is leveraged to derive a data-driven damage detection and localization method for wind turbine blades. In the proposed method, the blade's complex geometry is divided into less geometrically complex sections. Orthogonal strain maps are reconstructed from the sectioned hybrid dense sensor network by assuming different bidirectional shape functions and are solved using the least squares estimator. The error between the estimated strain maps and measured strains is extracted to define damage detection features that are dependent on the selected shape functions. This technique fuses sensor data into a single damage detection feature, providing a simple and robust method for inspecting large numbers of sensors without the need for complex model driven approaches. Numerical simulations demonstrate the proposed method's capability to distinguish healthy sections from possibly damaged sections on simplified 2D geometries.

Damage detection of shear buildings through structural mass-stiffness distribution

For structural damage detection of shear buildings, this paper proposes a new concept using structural element mass-stiffness vector (SEMV) based on special mass and stiffness distribution characteristics. A corresponding damage identification method is developed combining the SEMV with the cross-model cross-mode (CMCM) model updating algorithm. For a shear building, a model is assumed at the beginning based on the building's distribution characteristics. The model is updated into two models corresponding to the healthy and damaged conditions, respectively, using the CMCM method according to the modal parameters of actual structure identified from the measured acceleration signals. Subsequently, the structural SEMV for each condition can be calculated from the updated model using the corresponding stiffness and mass correction factors, and then is utilized to form a new feature vector in which each element is calculated by dividing one element of SEMV in health condition by the corresponding element of SEMV in damage condition. Thus this vector can be viewed as a damage detection feature for its ability to identify the mass or stiffness variation between the healthy and damaged conditions. Finally, a numerical simulation and the laboratory experimental data from a test-bed structure at the Los Alamos National Laboratory were analyzed to verify the effectiveness and reliability of the proposed method. Both simulated and experimental results show that the proposed approach is able to detect the presence of structural mass and stiffness variation and to quantify the level of such changes.

On the application of discrete-time Volterra series for the damage detection problem in initially nonlinear systems

Nonlinearities in the dynamical behavior of mechanical systems can degrade the performance of damage detection features based on a linearity assumption. In this article, a discrete Volterra model is used to monitor the prediction error of a reference model representing the healthy structure. This kind of model can separate the linear and nonlinear components of the response of a system. This property of the model is used to compare the consequences of assuming a nonlinear model during the nonlinear regime of a magneto-elastic system. Hypothesis tests are then employed to detect variations in the statistical properties of the damage features. After these analyses, conclusions are made about the application of Volterra series in damage detection.

PVDF Multielement Lamb Wave Sensor for Structural Health Monitoring.

The characteristics of Lamb waves, which are multimodal and dispersive, provide both challenges and opportunities for structural health monitoring (SHM). Methods for nondestructive testing with Lamb waves are well established. For example, mode content can be determined by moving a sensor to different positions and then transforming the spatial-temporal data into the wavenumber-frequency domain. This mode content information is very useful because at every frequency each mode has a unique wavestructure, which is largely responsible for its sensitivity to material damage. Furthermore, mode conversion occurs when the waves interact with damage, making mode content an excellent damage detection feature. However, in SHM, the transducers are typically at fixed locations and are immovable. Here, an affixed polyvinylidene fluoride (PVDF) multielement sensor is shown to provide these same capabilities. The PVDF sensor is bonded directly to the waveguide surface, conforms to curved surfaces, has low mass, low profile, low cost, and minimal influence on passing Lamb waves. While the mode receivability is dictated by the sensor being located on the surface of the waveguide, both symmetric and antisymmetric modes can be detected and group velocities measured.

A novel Bayesian imaging method for probabilistic delamination detection of composite materials

A probabilistic framework for location and size determination for delamination in carbon–carbon composites is proposed in this paper. A probability image of delaminated area using Lamb wave-based damage detection features is constructed with the Bayesian updating technique. First, the algorithm for the probabilistic delamination detection framework using the proposed Bayesian imaging method (BIM) is presented. Next, a fatigue testing setup for carbon–carbon composite coupons is described. The Lamb wave-based diagnostic signal is then interpreted and processed. Next, the obtained signal features are incorporated in the Bayesian imaging method for delamination size and location detection, as well as the corresponding uncertainty bounds prediction. The damage detection results using the proposed methodology are compared with x-ray images for verification and validation. Finally, some conclusions are drawn and suggestions made for future works based on the study presented in this paper.

Features for damage detection with insensitivity to environmental and operational variations

This paper explores and compares the application of three different approaches to the data normalization problem in structural health monitoring (SHM), which concerns the removal of confounding trends induced by varying operational conditions from a measured structural response that correlates with damage. The methodologies for singling out or creating damage-sensitive features that are insensitive to environmental influences explored here include cointegration, outlier analysis and an approach relying on principal component analysis. The application of cointegration is a new idea for SHM from the field of econometrics, and this is the first work in which it has been comprehensively applied to an SHM problem. Results when applying cointegration are compared with results from the more familiar outlier analysis and an approach that uses minor principal components. The ability of these methods for removing the effects of environmental/operational variations from damage-sensitive features is demonstrated and compared with benchmark data from the Brite-Euram project DAMASCOS (BE97 4213), which was collected from a Lamb-wave inspection of a composite panel subject to temperature variations in an environmental chamber.

Using vibration phase space topology changes for structural damage detection

Ideally, structural health monitoring of civil infrastructure consists of determining, by measured parameters, the location and severity of damage in the structure. Many structural vibration parameters have been used to identify and quantify damage. Using parameters based on structural vibration phase space features for damage detection is a new field in structural health monitoring. In this article, a new parameter based on topology changes of the phase space of vibration signals is proposed to identify structural damage, and an index named changes of phase space topology derived from vibration time history is used to locate the damage. A circular arch structure is used to demonstrate the method. Both numerical simulation and experimental tests of dynamic responses of a scaled arch structure to impact loads are carried out. The obtained structural response data are used to detect structural damage. Both the experimental and numerical results indicate that this method can successfully locate damage. It also demonstrated that this proposed method is more sensitive to damage but less sensitive to noise than modal-based parameters. The proposed damage index can be a good candidate in an online structural health monitoring system, as it depends on global vibration measurements but is more sensitive to structural damage than other global vibration-based parameters such as vibration frequencies and mode shapes.

Nonlinear features for damage detection on large civil structures due to earthquakes

This article proposes the degree of nonlinearity as a feature for damage detection on large civil structures due to earthquakes. The degree of nonlinearity, which is a numerical value, represents how nonlinear structures behave during earthquakes. The degree of nonlinearity is computed directly from the data of the ground motion and vibration of the structure. The computation of the degree of nonlinearity uses a Hilbert transform and does not require any other nonlinear structural or mathematical models. Therefore, the damage detection process is very fast. Thus, an immediate decision about the condition of structures after strong earthquakes is possible. This article shows theoretically that a plot between the degree of nonlinearity against the magnitude of the ground motion represents the behavior of a structure. This plot is considered the signature of that structure. A structure that is healthy will have its specific healthy signature. If a new earthquake strikes and the new degree of nonlinearity deviates from its healthy signature, the structure may be damaged by that earthquake. To verify the validity of the degree of nonlinearity, numerical studies of nonlinear systems are presented. Although the application of degree of nonlinearity on a real civil structure is still not totally conclusive due to the limitation of data, the results from currently available data are encouraging. With more investigation, the degree of nonlinearity could be an additional or alternative feature for vibration-based structural health monitoring.

Lamb Wave-Based Damage Location Detection Method of Composite Plate

To understand more about Lamb waves on composite laminates damage detection features, the Lamb wave group velocity dispersion curves are calculated and plotted by using dichotomy method in MATLAB. The signal parameters are chosen according to Group velocity dispersion curves. The dynamic response signals of the composite plate are obtained by finite element method. Damage location is calculated by the actual group velocity of Lamb wave and time of flight of the difference signal before and after damage.

Experimental validation of a new statistical process control feature for damage detection

A novel statistical process control-based method for damage detection in cantilever-like structures is presented in this paper . It is based on time-domain data obtained from the free vibrations of the first mode of the structure, measured at only one point. The concept of signal length is introduced here as a feature for statistical process control. A statistical model consistent with extreme value theory is fitted to the operational data and used to establish control limits. A strategy based on controlling runs is adopted to increase the sensitivity of the detection. The procedure was applied to a four-storey steel frame, which was subjected in turn to mass and stiffness modifications by adding masses or loosening bolts in a beam–column connection. The results were successful; the proposed procedure was able to detect an increase of 0.04% of the total mass of the frame and loosening of one out of 152 bolts.

Bifurcation boundary analysis as a nonlinear damage detection feature: does it work?

Many systems in engineering and science are inherently nonlinear and require damage detection. For such systems, nonlinear damage detection methods may be useful. A bifurcation boundary analysis method as a new nonlinear damage detection tool was previously introduced in the literature to track bifurcation boundary changes due to damages over a small region of an aeroelastic panel model. Results of this method based upon a finite difference solution showed higher sensitivities to the small amount of damage than methods based upon linear models. In this paper, four methods including Finite Difference, Finite Element (FEM), Rayleigh–Ritz and Galerkin Approach are used to further investigate the sensitivity of the bifurcation boundary for damage detection. Results of the FEM and Rayleigh–Ritz method agree with each other and also show that the sensitivity of the bifurcation boundary to damage is much less than what previously reported when using a finite difference solution method.