Online Structural Health Monitoring Research Articles

Online structural health monitoring of polymer composite structure using gold nanoparticles (AuNPs)

A mechanism for online structural health monitoring of composite structure was established by developing highly sensitive strain gauges based on colloidal Gold Nano Particles (AuNPs). They were synthesized using well known Turkevich-Frens method followed by phase transfer in chloroform using surfactant Cetyltrimethylammonium bromide (CTAB). Flexible Strain gauges / smart sensors were formed by mixing colloidal AuNPs-chloroform solution in polystyrene (PS) – chloroform solution and depositing their thick paste on hand-made flexible thermoplastic polyurethane (TPU) film. These sensors were used as strain gauges on composite substrates for tensile testing. The developed AuNP based strain gauges were found to have 10 times higher sensitivity than normal metal foil strain gauges.

A Stiffness Approach for Coupling Structural and Magnetic Models for the Sustainable Design, Optimisation and Real-Time Structural Integrity Assessment of Radial Flux Permanent Magnet Generators for Direct-Drive Wind Turbines

The mass of a direct-drive generator is often defined by the requirements for structural stiffness to meet the magnetic stiffness between the rotor and stator surfaces. This paper analyses this magnetic stiffness and estimates the structural stiffness of direct-drive generators for different modes of deflection. The magnetic stiffness modelling is based on an analytical model of the airgap closing forces. The final models are verified using finite element analysis and developed for both permanent magnet and wound rotor generators. It shows that wound rotor machines have higher stiffness requirements than permanent magnet machines. The structural stiffness of the generator rotor and stator is evaluated for different modes by applying spatially varying forces and finding the associated deflections. Structural stiffnesses for the rotor, stator and bearing are then combined. Finally, the magnetic and structural stiffnesses are combined and a stiffness margin can be found. This method is applied to a relatively stiff and a relatively compliant set of generator structures in a case study. The analytical model presented in this paper is useful for structural optimisation purposes or as part of an online structural health monitoring system as it could assess the integrity of the machines in real time.

Development of Recursive Subspace Identification for Real-Time Structural Health Monitoring under Seismic Loading

Structural health monitoring (SHM) can continuously and nondestructively evaluate the state and performance of structures using the structural responses to external loads or environmental conditions. Moreover, online or real-time SHM of civil structures provides significant advantages over periodic or manual inspection methods, especially under disaster loadings, where the consequences of failure can be severe. To achieve it, performing system identification and damage detection recursively, said recursive subspace identification (RSI), is a promising solution, and SHM based on the algorithms can evaluate damage or deterioration of civil structures, give insight into the health and performance of a structural system, and provide valuable information for decision-making on maintenance and repair. However, the time-consuming decompositions frustrate these algorithms. As a compromise, additional processing is required to implement online and real-time applications. This study demonstrates a modified algorithm that takes advantage of the projection approximation subspace tracking (PAST) algorithm and the repeated system matrices in the extended observability matrix. The modification can reduce numerical decompositions and improve important timeliness for online or real-time SHM of civil structures. Both the numerical simulation and experimental investigation have been used to verify the proposed method, and the results show its capability to determine the changes in the dynamic characteristics of a structure in either the laboratory experiment or in the field application. In the last place, the discussion and some conclusions are also drawn in this paper.

Output-only modal identification with recursive dynamic mode decomposition for time-varying systems

The time-varying modal identification is a prominent analytical means but challenging task to grasp the dynamic characteristics of time-varying systems, which is important guarantee for real-time control and on-line health monitoring of many engineering structures. In this paper, a data-driven approach for time-varying modal identification without requiring parametric modeling about underlying system is presented. Firstly, the fundamental theory for Dynamic Mode Decomposition (DMD) is reviewed combining the Koopman operator. Subsequently, the DMD incorporated into operational modal identification is examined from the perspective of linear system, and the inner connection between modal parameters of vibration system and eigenvalues and eigenmodes of DMD is established. On the basis of that, a weighted recursive strategy integrated with DMD is employed to investigate the capability of time-varying modal identification, and the main identification procedures are elaborated. Finally, the effectiveness and practicability of the proposed method in extracting modal properties of time-varying systems are investigated by a numerical and experiment example. The results demonstrate that, compared to other methods, the proposed approach has high computational efficiency while ensuring the identification accuracy, exhibiting a good engineering application prospect.

Bayesian incremental learning paradigm for online monitoring of dam behavior considering global uncertainty

Advanced data-driven methods for real-time monitoring of dam structural behavior, especially displacement, are obtaining increasing popularity. Most existing data-driven models are offline and static, however, which cannot be dynamically updated with the increase of monitoring data. Additionally, the capability of monitoring models to identify anomalies, which is the ultimate goal of dam behavior modeling, has received little attention. This study proposes a novel Extreme Learning Machine (ELM)-based Bayesian incremental learning methodology to address these challenges. The proposed model is termed Incremental Global Bayesian ELM (I-GBELM), which couples ELM with global Bayesian and incremental learning for online monitoring of dam displacements. Firstly, a global Bayesian approach to the ELM is presented to account for both input and target uncertainties and to infer the network parameters analytically. Secondly, an incremental learning paradigm is theoretically derived, which aims to directly learn new knowledge from additional data to update model parameters without retraining from scratch. The I-GBELM model integrates the strengths of both methods so that it allows accurately predicting displacements, identifying anomalies, and efficiently updating parameters. For online deployment purposes, finally, a non-iterative pruning algorithm is developed to adaptively discard the redundant hidden nodes, yielding a more compact model. The performance of the proposed model is fully validated through comprehensive verification cases using benchmark and real-world monitoring datasets. The empirical results indicate that the proposed I-GBELM outperforms all other models in most cases, and hence, is a competitive modeling tool for online structural health monitoring.

Prognostics-based adaptive control strategy for lifetime control of wind turbines

Abstract. Variability in wind profiles in both space and time is responsible for fatigue loading in wind turbine components. Advanced control methods for mitigating structural loading in these components have been proposed in previous works. These also incorporate other objectives like speed and power regulation for above-rated wind speed operation. In recent years, lifetime control and extension strategies have been proposed to guarantee power supply and operational reliability of wind turbines. These control strategies typically rely on a fatigue load evaluation criteria to determine the consumed lifetime of these components, subsequently varying the control set point to guarantee a desired lifetime of the components. Most of these methods focus on controlling the lifetime of specific structural components of a wind turbine, typically the rotor blade or tower. Additionally, controllers are often designed to be valid about specific operating points and hence exhibit deteriorating performance in varying operating conditions. Therefore, they are not able to guarantee a desired lifetime in varying wind conditions. In this paper an adaptive lifetime control strategy is proposed for controlled aging of rotor blades to guarantee a desired lifetime while considering damage accumulation level in the tower. The method relies on an online structural health monitoring system to vary the lifetime controller gains based on a state-of-health (SoH) measure by considering the desired lifetime at every time step. For demonstration, a 1.5 MW National Renewable Energy Laboratory (NREL) reference wind turbine is used. The proposed adaptive lifetime controller regulates structural loading in the rotor blades to guarantee a predefined damage level at the desired lifetime without sacrificing the speed regulation performance of the wind turbine. Additionally, a significant reduction in the tower fatigue damage is observed.

An Improved Prototype Network and Data Augmentation Algorithm for Few-Shot Structural Health Monitoring Using Guided Waves

The significance of implementing online structural health monitoring (SHM) for aerospace structures under harsh service environments cannot be overemphasized. Deep learning has demonstrated a promising and effective means to achieve accommodate such a need. However, as envisaged, the performance of deep learning-facilitated SHM heavily relies on the scale of training dataset, degrading the practicability of the approach. To address this, we propose an improved prototype network and data augmentation methods for few-shot SHM using guided waves. In the improved prototype network, the weighted Euclidean distance is used for damage classification. An attention module is established to predict the weight coefficients. The Davies–Bouldin Index (DBI) is used in the loss function to better separate the embedding vectors of different classes. Time masking and frequency masking are proposed for data augmentation of guided wave signals. As bolt joints are widely used in aerospace structures, the proposed approach is experimentally validated by quantifying the degree of bolt loosening in multibolt connection structures. The results are compared against those obtained from other classical few-shot learning (FSL) methods.

Operational modal analysis with compressed measurements based on prior information

Compressive Sensing (CS) provides a solution for modal analysis with fewer/compressed measurements. However, operational modal analysis by using compressed measurements directly has not been studied. In this work, a new approach is proposed, which deploys the identified modal parameters from the most recent step as prior information to automatically set the frequency and damping ratio search ranges and uses CS to identify modal parameters from compressed measurements. Numerical studies on a spring-mass system and experimental studies on a wind turbine model are performed to demonstrate the effectiveness of the proposed approach. The results are accurate on damped structures including those with heavy noise and complex responses, under free or random vibration conditions. Its performance using compressed measurements is comparable to those of state-of-the-art methods using original uncompressed measurements. It may find practical applications as an efficient and accurate operational modal analysis tool for online structural health monitoring.

Automated operational modal analysis using variational Gaussian mixture model

Automated operational modal analysis is essential for online structural health monitoring without human intervention. It remains a challenging issue due to the need of processing a large number of datasets and the involvement of many user-specified thresholds. This paper proposes a novel automated modal identification approach based on stochastic subspace identification and variational Gaussian mixture model that involves the analysis of the stabilization diagram to automatically identify modal parameters. Two validation criteria are first adopted to eliminate the spurious modes in the stabilization diagram. A Gaussian mixture model (GMM) is then used to probabilistically classify each pole in the stabilization diagram to a specific cluster. The parameters of GMM are estimated using variational inference, giving representatives of each mode, and the optimal number of clusters is automatically determined through the Dirichlet Process. The proposed framework automatically distinguishes physical modes from spurious modes with only two verified and widely used thresholds. Results of a four-story shear and a footbridge with continuous measurements demonstrate the efficacy and robustness of the proposed approach. It shows that the proposed approach can automatically identify modal parameters with high accuracy, including weakly excited and closely spaced modes.

Detection of the instantaneous frequency degradation due to damages of a fixed offshore jacket platform using the iEEMD‐based Hilbert Huang transform under a wave excitation

In online structural health monitoring of the fixed offshore jacket platform, the determination of the time-dependent degradation of structures due to damages is the most important step in a maintenance strategy. This paper presents the application of the Hilbert Huang transform method based on the iEEMD (improved ensemble empirical mode decomposition) algorithm to identify the instantaneous natural frequency degradation due to damages of the fixed offshore structure under the random wave excitation. The iEEMD algorithm is proposed to restrain the mode mixing and used to adaptively process the big response data of the offshore structure during the wave loading with a wide frequency range. The vibration response measurement experiment for the offshore structure was conducted in a wave tank in this study. The identified fundamental natural frequencies by the iEEMD-based Hilbert Huang method were evaluated and compared with results of the numerical model of the eigenvalue analysis. The main result demonstrates that the instantaneous frequency degradation and nonlinear time-varying behavior due to the brace damage of the fixed offshore jacket platform can be successfully identified by the iEEMD-based Hilbert Huang transform technique under a wave load.

An Energy-Based Damage Diagnosis Under Changing Environmental Temperature for Online SHM

Structural health monitoring (SHM) for early diagnosis of damage in structures has become vital in improving the structural integrity and safety of the structure and thereby providing a smooth operation for most of the important in-service civil structures such as buildings and bridges. SHM also contributes to the substantial reduction in the maintenance cost of the structure. During serious structural failures, SHM can safeguard human life through an early warning system. This paper presents an output-only model-free technique by extracting damage features from the energy profiles derived from measured time-history responses. The proposed algorithm is organized into two phases i.e. the learning phase and the monitoring phase. While the learning phase can optionally be executed offline, the monitoring phase can preferably be online. The damage feature is extracted by carrying out a fractal dimension analysis of the continuous waveform of energy associated with the jolt (i.e. rate of change of acceleration) evaluated from the acceleration responses of the structure. A novel damage index is proposed, to localize the damage, using the damage features extracted from the fractal dimension, while handling operational and environmental variability (EoV). The proposed algorithm is evaluated through numerical simulations on a typical span of 220[Formula: see text]m RCC bridge across river Amaravathi in Tamil Nadu, India and also a 20-story frame structure. Experimental investigations are also carried out by testing a laboratory-scale 10-story frame structure and also an RCC bridge model of a 7.7[Formula: see text]m span to evaluate the practical amenability of the proposed algorithm. The experimental and numerical studies carried out in this paper, suggest that the proposed energy-based diagnostic technique can effectively localize the damage even in the presence of confounding factors like EoV. Conclusively, the proposed diagnostic technique can be an effective tool for online SHM, as it is model-free with less tedious computations and can perform well even in the presence of noise.

Digital image correlation (DIC) based damage detection for CFRP laminates by using machine learning based image semantic segmentation

Vision-based damage detection in carbon fiber-reinforced plastic (CFRP) composites can be interfered by such factors as surface texture, stains and lighting. A digital image correlation (DIC) based surface strain monitoring technique, on the other hand, enables to track the change of strain distribution. It is promising to develop a new approach for online structural health monitoring (SHM), in which the DIC strain contours can be scrutinized automatically and the results are no longer substantially subjected to human interference. In this study, a convolutional neural network (CNN) based image semantic segmentation technique is proposed for pixel-level classification of DIC strain field images. A DeepLabv3+ encoder-decoder architecture combined with different feature extraction networks is investigated. The training dataset and validation of the model are obtained through finite element (FE) simulation. The images of quasi-static axial tensile strain field obtained from 2D-DIC are used to test the accuracy and efficiency of the trained CNN model. It is found that use of a pre-trained ResNet-50 CNN model as the backbone network of DeepLabv3+ architecture through a transfer learning algorithm can make the semantic segmentation results reach a mean intersection over union of 0.9236. The prediction accuracy of the semantic segmentation model trained from the FE data is comparable with that of the model trained from the experimental data, which demonstrates that the proposed machine learning approach for DIC measurement is cost-effective.

Study of Ultra High Frequency Measurement Techniques for Online Monitoring of Partial Discharges in High Voltage Systems

Partial Discharge (PD) activity is a pre-cursor for insulation degradation which may eventually lead to catastrophic failure of the electrical equipment with severe social and economic consequences. It is therefore imperative that PD is detected at its early stages to permit repair or replacement, prior to total failure. The bulky non-planar construction of existing ultra-high frequency (UHF) sensors coupled with the lack of proper feeding techniques and the ultrawide signal bandwidth of PD signals which necessitates the use of digital storage oscilloscope have restricted field deployment of UHF technique for online PD detection. In this work, we present a compact ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.58 \boldsymbol {\lambda } $ </tex-math></inline-formula> ) wideband (0.5-3 GHz) dual arm Archimedean slot spiral with coplanar waveguide feed for PD detection, where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\boldsymbol {\lambda } $ </tex-math></inline-formula> is the wavelength at center frequency. Numerical simulations of the antenna were compared with the measurements of the fabricated antenna. The fabricated wideband antenna was used to detect PD emissions from a mineral oil test cell connected to 10 kVA 50 Hz transformer. We also present different UHF signal measurement techniques ranging from PD pulse detection to reconstruction of the bandlimited PD signal in the digital domain for online structural health monitoring of high voltage systems. The proposed UHF PD measurement techniques are low cost compared to the conventional approach and indicate the feasibility of an embedded system for online PD monitoring using the fabricated UHF sensor.

Low-velocity impact source localization in a composite sandwich structure using a broadband piezoelectric sensor network

An online structural health monitoring strategy for real-time impact source localisation in composite sandwich structure (CSS) is presented in this paper. A broadband piezoelectric (PZT) sensor-network based laboratory experiments and finite element method based numerical simulations of low-velocity steel-ball drop impacts at different locations on a CSS have been carried out. It is observed that the arrival-time of the impact-induced elastic wave signals increases with the increase in source-to-sensor distance. Further, the influence of impactor drop-height variations is studied by conducting a series of numerical simulations in ABAQUS. An arrival-time difference calculation based impact source localisation strategy is proposed to locate the impact regions in the CSS. The proposed monitoring strategy uses the registered impact signals (time-domain) at the PZTs to predict impact locations in real-time. Eventually, a study on the estimation of localisation error has been carried out to check the effectiveness of the proposed SHM strategy.

The Instantaneous Spectral Entropy for Real-time, Online Structural Health Monitoring.

Abstract Entropy measurements have been recently proposed for the damage assessment of civil structures and mechanical systems. Here, a quasi-real-time approach, based on Instantaneous Spectral Entropy (ISE), is proposed for the detection of sudden stiffness reduction, breathing cracks, and other kinds of structural changes. The method, validated on an experimental benchmark, is suitable for nonstationary signals originating from nonlinear structures as well.

Numerical study on the displacement reconstruction of subsea pipelines using the improved inverse finite element method

The structural health monitoring (SHM) and integrity management of subsea pipelines are essential in the offshore oil and gas industry. Previous studies have achieved strain and thermal monitoring using distributed fiber optic sensors (DFOS). However, assessing the in situ displacement of subsea pipelines is still challenging. This study put forward a displacement reconstruction approach based on limited strain data using the inverse finite element method (iFEM). First, we developed a new inverse beam element with a 2nd order shape function for axial deformation to extend the application scope of the standard iFEM to subsea pipelines undergoing complex axial forces. Then, a model truncation procedure and a continuous displacement reconstruction algorithm were proposed to rebuild high accuracy and smooth displacement field. Besides, the strain sensing scheme for iFEM was adapted to DFOS system and made easier to implement in engineering applications. Finally, the proposed method was verified through a series of numerical studies. It showed that the improved iFEM offers superior accuracy and better noise robustness than the standard iFEM, especially with coarse meshes and low spatial resolution strain data. Thus, it reveals a promising prospect in the online displacement monitoring and SHM for subsea pipelines.

Toward a general unsupervised novelty detection framework in structural health monitoring

AbstractThis study proposes an unsupervised, online structural health monitoring framework robust to the sensor configuration, that is, the number and placement of sensors. The proposed methodology leverages generative adversarial networks (GANs). The GAN's discriminator network is the novelty detector, while its generator provides additional data to tune the detection threshold. GAN models are trained with the fast Fourier transform of structural accelerations as input, avoiding the need for any structure‐specific feature extraction. Dense, convolutional (convolutional neural network), and long short‐term memory (LSTM) units are evaluated as discriminators under different GAN training loss patterns, that is, the differences between discriminator and generator training losses. Results show that the LSTM‐based discriminators and the suggested threshold tuning technique to be robust even in loss patterns with overfitted discriminators, a probable outcome of limited training sets. The framework is evaluated on two benchmark datasets. With only 100 s of training data, it achieved 95% novelty detection accuracy, distinguishing between different damage classes and identifying their resurgence under varying sensor configurations. Finally, the majority‐vote‐ensemble of discriminator‐generator pairs at different training epochs is introduced to reduce false alarms, improve novelty detection accuracy and stability.

CNN and Convolutional Autoencoder (CAE) based real-time sensor fault detection, localization, and correction

Increasing advances in sensing technologies and analytics have led to the proliferation of sensors to monitor structural and infrastructural systems. Accurate sensor data can provide information about structural health, aid in prognosis, and help calculate forces for vibration control. However, sensors are susceptible to faults such as loss of data, random noise, bias, drift, etc., due to the aging of sensors, defects, or environmental factors. Although traditional signal processing techniques can detect and isolate faults and reconstruct corrupt or missing sensor data, they demand significant human intervention. The continuous rise in computational power and demonstrated efficacy in numerous domains motivates the use of deep learning to minimize human-in-the-loop techniques. In this work, we introduce a novel, deep learning framework for linear systems with time-invariant parameters that identifies the presence and type of fault in sensor data, location of the faulty sensor and subsequently reconstructs the correct sensor data for fault detection, fault classification, and reconstruction. In our framework, first, a Convolutional Neural Network (CNN) is used to detect the presence of a fault and identify its type. Next, a suite of individually trained Convolutional Autoencoder (CAE) networks corresponding to each type of fault are employed for reconstruction. We demonstrate the efficacy of our framework to address both single and multiple sensor faults in synthetically generated data of a simple shear-type structure and experimentally measured data from a simplified arch bridge. While the framework is agnostic of fault-type, we demonstrate its use for four types of fault namely, missing, spiky, random, and drift. For both simulated and experimental datasets with a single fault, our models performed well, achieving 100% accuracy in faulty sensor localization, more than 98.7% accuracy in fault type detection, and more than 99% accuracy in reconstruction. Our framework can also address multiple concurrent faults with similar accuracy. We empirically demonstrate that our proposed framework performs better than other state-of-the-art techniques in terms of computational efficiency with comparable accuracy. Adoption of our framework in online structural health monitoring applications can lead to minimal disruption to monitoring processes, reduced downtime for structures and infrastructure while simultaneously reducing uncertainty and improving the quality of sensor data for historical records.

Online structural integrity monitoring of high-performance composite rocket motor casing

Solid rocket motor (SRM) is one of the significant sub-systems for aerospace launch vehicles. The requirements of high range and high payload capabilities in aerospace systems demand the usage of high specific strength and high specific stiffness materials for rocket motor casings to minimize the inert weights. The present state-of-the-art technology is to realize stages of missile systems/launch vehicles with Composite Rocket Motor Casing (CRMC) to have a high-performance factor. CRMC is being designed with a marginal Factor of Safety (FOS) and mostly for a one-time use during its service life. CRMC is pressure tested to verify leak containment and its ability to sustain the Maximum Expected Operating Pressure (MEOP). Acoustic Emission Testing (AET), an online and global Structural Health Monitoring (SHM) technique along with strain and dilation measurement is used for observing the dynamic behavior of CRMC during the pressure test. Conventional Non-destructive Testing (NDT) such as Ultrasonic (UT) and Radiographic testing (RT) are employed to identify the type of defects. This unique capability of monitoring the dynamic behavior of active defects during real-time is of great significance for the acceptance of CRMC. The evolution of such an online system requires, first, identification of an appropriate process and next its adoption to meet the challenges posed by composite materials. There are various steps involved in the online SHM of CRMC. In this paper, various test practices, requirements, test parameters along with the details of strain and dilation measurements are given. This online inspection methodology and strain-dilation measurement during the Proof Pressure Test (PPT), along with a series of other inspection methods form the basis of CRMC acceptance for further use.

Tracking of Stiffness Variation in Structural Members Using Input Error Function Observers

This study evaluates input error function observers for tracking of stiffness variation in real-time. The input error function is an Analytical Redundancy (AR)-based diagnosis method and necessitates a mathematical model of the system and system identification techniques. In practice, mathematical models used during numerical simulations differ from the actual status of the structure, and thus, accurate mathematical models are rarely available for reference. Noise is an unwanted signal in the input–output measurements but unavoidable in real-world applications (as in long span bridge trusses) and hard to imitate during numerical simulations. Simulation data from the truss system clearly indicates the effectiveness of the proposed structural damage detection method for estimating the severity of the damage. Optimization of the input error function can further automate the stiffness estimation in structural members and address critical aspects such as system uncertainties and the presence of noise in input–output measurements. Stiffness tracking in one of the planar truss members indicates the potential of optimization of the input error function for online structural health monitoring and implementing condition-based maintenance.