Fusing Expert Knowledge with Monitoring Data for Condition Assessment of Railway Welds.

Monitoring information can facilitate the condition assessment of railway infrastructure, via delivery of data that is informative on condition. A primary instance of such data is found in Axle Box Accelerations (ABAs), which track the dynamic vehicle/track interaction. Such sensors have been installed on specialized monitoring trains, as well as on in-service On-Board Monitoring (OBM) vehicles across Europe, enabling a continuous assessment of railway track condition. However, ABA measurements come with uncertainties that stem from noise corrupt data and the non-linear rail-wheel contact dynamics, as well as variations in environmental and operational conditions. These uncertainties pose a challenge for the condition assessment of rail welds through existing assessment tools. In this work, we use expert feedback as a complementary information source, which allows the narrowing down of these uncertainties, and, ultimately, refines assessment. Over the past year, with the support of the Swiss Federal Railways (SBB), we have assembled a database of expert evaluations on the condition of rail weld samples that have been diagnosed as critical via ABA monitoring. In this work, we fuse features derived from the ABA data with expert feedback, in order to refine defection of faulty (defect) welds. Three models are employed to this end; Binary Classification and Random Forest (RF) models, as well as a Bayesian Logistic Regression (BLR) scheme. The RF and BLR models proved superior to the Binary Classification model, while the BLR model further delivered a probability of prediction, quantifying the confidence we might attribute to the assigned labels. We explain that the classification task necessarily suffers high uncertainty, which is a result of faulty ground truth labels, and explain the value of continuously tracking the weld condition.

Dynamic Response Study of Piezoresistive Ti3C2-MXene Sensor for Structural Impacts.

MXenes are a new family of two-dimensional (2D) nanomaterials. They are inorganic compounds of metal carbides/nitrides/carbonitrides. Titanium carbide MXene (Ti3C2-MXene) was the first 2D nanomaterial reported in the MXene family in 2011. Owing to the good physical properties of Ti3C2-MXenes (e.g., conductivity, hydrophilicity, film-forming ability, elasticity) various applications in wearable sensors, energy harvesters, supercapacitors, electronic devices, etc., have been demonstrated. This paper presents the development of a piezoresistive Ti3C2-MXene sensor followed by experimental investigations of its dynamic response behavior when subjected to structural impacts. For the experimental investigations, an inclined ball impact test setup is constructed. Stainless steel balls of different masses and radii are used to apply repeatable impacts on a vertical cantilever plate. The Ti3C2-MXene sensor is attached to this cantilever plate along with a commercial piezoceramic sensor, and their responses for the structural impacts are compared. It is observed from the experiments that the average response times of the Ti3C2-MXene sensor and piezoceramic sensor are 1.28±0.24µs and 31.19±24.61µs, respectively. The fast response time of the Ti3C2-MXene sensor makes it a promising candidate for monitoring structural impacts.

Remaining Useful Life Estimation for Engineered Systems Operating under Uncertainty with Causal GraphNets.

In this work, a novel approach, termed GNN-tCNN, is presented for the construction and training of Remaining Useful Life (RUL) models. The method exploits Graph Neural Networks (GNNs) and deals with the problem of efficiently learning from time series with non-equidistant observations, which may span multiple temporal scales. The efficacy of the method is demonstrated on a simulated stochastic degradation dataset and on a real-world accelerated life testing dataset for ball-bearings. The proposed method learns a model that describes the evolution of the system implicitly rather than at the raw observation level and is based on message-passing neural networks, which encode the irregularly sampled causal structure. The proposed approach is compared to a recurrent network with a temporal convolutional feature extractor head (LSTM-tCNN), which forms a viable alternative for the problem considered. Finally, by taking advantage of recent advances in the computation of reparametrization gradients for learning probability distributions, a simple, yet efficient, technique is employed for representing prediction uncertainty as a gamma distribution over RUL predictions.

Kalman Filter-Based Fusion of Collocated Acceleration, GNSS and Rotation Data for 6C Motion Tracking.

The ground motion of an earthquake or the ambient motion of a large engineered structure not only has translational motion, but it also includes rotation around all three axes. No current sensor can record all six components, while the fusion of individual instruments that could provide such recordings, such as accelerometers or Global Navigation Satellite System (GNSS) receivers, and rotational sensors, is non-trivial. We propose achieving such a fusion via a six-component (6C) Kalman filter (KF) that is suitable for structural monitoring applications, as well as earthquake monitoring. In order to develop and validate this methodology, we have set up an experimental case study, relying on the use of an industrial six-axis robot arm, on which the instruments are mounted. The robot simulates the structural motion resulting atop a wind-excited wind turbine tower. The quality of the 6C KF reconstruction is assessed by comparing the estimated response to the feedback system of the robot, which performed the experiments. The fusion of rotational information yields significant improvement for both the acceleration recordings but also the GNSS positions, as evidenced via the substantial reduction of the RMSE, expressed as the difference between the KF predictions and robot feedback. This work puts forth, for the first time, a KF-based fusion for all six motion components, validated against a high-precision ground truth measurement. The proposed filter formulation is able to exploit the strengths of each instrument and recover more precise motion estimates that can be exploited for multiple purposes.

A Novel Approach for 3D-Structural Identification through Video Recording: Magnified Tracking.

Advancements in optical imaging devices and computer vision algorithms allow the exploration of novel diagnostic techniques for use within engineering systems. A recent field of application lies in the adoption of such devices for non-contact vibrational response recordings of structures, allowing high spatial density measurements without the burden of heavy cabling associated with conventional technologies. This, however, is not a straightforward task due to the typically low-amplitude displacement response of structures under ambient operational conditions. A novel framework, namely Magnified Tracking (MT), is proposed herein to overcome this limitation through the synergistic use of two computer vision techniques. The recently proposed phase-based motion magnification (PBMM) framework, for amplifying motion in a video within a defined frequency band, is coupled with motion tracking by means of particle tracking velocimetry (PTV). An experimental campaign was conducted to validate a proof-of-concept, where the dynamic response of a shear frame was measured both by conventional sensors as well as a video camera setup, and cross-compared to prove the feasibility of the proposed non-contact approach. The methodology was explored both in 2D and 3D configurations, with PTV revealing a powerful tool for the measurement of perceptible motion. When MT is utilized for tracking "imperceptible" structural responses (i.e., below PTV sensitivity), via the use of PBMM around the resonant frequencies of the structure, the amplified motion reveals the operational deflection shapes, which are otherwise intractable. The modal results extracted from the magnified videos, using PTV, demonstrate MT to be a viable non-contact alternative for 3D modal identification with the benefit of a spatially dense measurement grid.

Cost⁻Benefit Optimization of Structural Health Monitoring Sensor Networks.

Structural health monitoring (SHM) allows the acquisition of information on the structural integrity of any mechanical system by processing data, measured through a set of sensors, in order to estimate relevant mechanical parameters and indicators of performance. Herein we present a method to perform the cost⁻benefit optimization of a sensor network by defining the density, type, and positioning of the sensors to be deployed. The effectiveness (benefit) of an SHM system may be quantified by means of information theory, namely through the expected Shannon information gain provided by the measured data, which allows the inherent uncertainties of the experimental process (i.e., those associated with the prediction error and the parameters to be estimated) to be accounted for. In order to evaluate the computationally expensive Monte Carlo estimator of the objective function, a framework comprising surrogate models (polynomial chaos expansion), model order reduction methods (principal component analysis), and stochastic optimization methods is introduced. Two optimization strategies are proposed: the maximization of the information provided by the measured data, given the technological, identifiability, and budgetary constraints; and the maximization of the information⁻cost ratio. The application of the framework to a large-scale structural problem, the Pirelli tower in Milan, is presented, and the two comprehensive optimization methods are compared.

A Data-Driven Diagnostic Framework for Wind Turbine Structures: A Holistic Approach.

The complex dynamics of operational wind turbine (WT) structures challenges the applicability of existing structural health monitoring (SHM) strategies for condition assessment. At the center of Europe's renewable energy strategic planning, WT systems call for implementation of strategies that may describe the WT behavior in its complete operational spectrum. The framework proposed in this paper relies on the symbiotic treatment of acting environmental/operational variables and the monitored vibration response of the structure. The approach aims at accurate simulation of the temporal variability characterizing the WT dynamics, and subsequently at the tracking of the evolution of this variability in a longer-term horizon. The bi-component analysis tool is applied on long-term data, collected as part of continuous monitoring campaigns on two actual operating WT structures located in different sites in Germany. The obtained data-driven structural models verify the potential of the proposed strategy for development of an automated SHM diagnostic tool.

VpROM: a novel variational autoencoder-boosted reduced order model for the treatment of parametric dependencies in nonlinear systems.

Reduced Order Models (ROMs) are of considerable importance in many areas of engineering in which computational time presents difficulties. Established approaches employ projection-based reduction, such as Proper Orthogonal Decomposition. The limitation of the linear nature of such operators is typically tackled via a library of local reduction subspaces, which requires the assembly of numerous local ROMs to address parametric dependencies. Our work attempts to define a more generalisable mapping between parametric inputs and reduced bases for the purpose of generative modeling. We propose the use of Variational Autoencoders (VAEs) in place of the typically utilised clustering or interpolation operations, for inferring the fundamental vectors, termed as modes, which approximate the manifold of the model response for any and each parametric input state. The derived ROM still relies on projection bases, built on the basis of full-order model simulations, thus retaining the imprinted physical connotation. However, it additionally exploits a matrix of coefficients that relates each local sample response and dynamics to the global phenomena across the parametric input domain. The VAE scheme is utilised for approximating these coefficients for any input state. This coupling leads to a high-precision low-order representation, which is particularly suited for problems where model dependencies or excitation traits cause the dynamic behavior to span multiple response regimes. Moreover, the probabilistic treatment of the VAE representation allows for uncertainty quantification on the reduction bases, which may then be propagated to the ROM response. The performance of the proposed approach is validated on an open-source simulation benchmark featuring hysteresis and multi-parametric dependencies, and on a large-scale wind turbine tower characterised by nonlinear material behavior and model uncertainty.

Seismic assessment of bridges through structural health monitoring: a state-of-the-art review.

The present work offers a comprehensive overview of methods related to condition assessment of bridges through Structural Health Monitoring (SHM) procedures, with a particular interest on aspects of seismic assessment. Established techniques pertaining to different levels of the SHM hierarchy, reflecting increasing detail and complexity, are first outlined. A significant portion of this review work is then devoted to the overview of computational intelligence schemes across various aspects of bridge condition assessment, including sensor placement and health tracking. The paper concludes with illustrative examples of two long-span suspension bridges, in which several instrumentation aspects and assessments of seismic response issues are discussed.

Symplectic encoders for physics-constrained variational dynamics inference.

We propose a new variational autoencoder (VAE) with physical constraints capable of learning the dynamics of Multiple Degree of Freedom (MDOF) dynamic systems. Standard variational autoencoders place greater emphasis on compression than interpretability regarding the learned latent space. We propose a new type of encoder, based on the recently developed Hamiltonian Neural Networks, to impose symplectic constraints on the inferred a posteriori distribution. In addition to delivering robust trajectory predictions under noisy conditions, our model is capable of learning an energy-preserving latent representation of the system. This offers new perspectives for the application of physics-informed neural networks on engineering problems linked to dynamics.

Octet lattice-based plate for elastic wave control.

Motivated by the importance of lattice structures in multiple fields, we numerically investigate the propagation of flexural waves in a thin reticulated plate augmented with two classes of metastructures for wave mitigation and guiding, namely metabarriers and metalenses. The cellular architecture of this plate invokes the well-known octet topology, while the metadevices rely on novel customized octets either comprising spherical masses added to the midpoint of their struts or variable node thickness. We numerically determine the dispersion curves of a doubly-periodic array of octets, which produce a broad bandgap whose underlying physics is elucidated and leveraged as a design paradigm, allowing the construction of a metabarrier effective for inhibiting the transmission of waves. More sophisticated effects emerge upon parametric analyses of the added masses and node thickness, leading to graded designs that spatially filter waves through an enlarged bandgap via rainbow trapping. Additionally, Luneburg and Maxwell metalenses are realized using the spatial modulation of the tuning parameters and numerically tested. Wavefronts impinging on these structures are progressively curved within the inhomogeneous media and steered toward a focal point. Our results yield new perspectives for the use of octet-like lattices, paving the way for promising applications in vibration isolation and energy focusing.

Vibration-based monitoring of a small-scale wind turbine blade under varying climate conditions. Part I: An experimental benchmark.

Structural health monitoring (SHM) has been increasingly exploited in recent years as a valuable tool for assessing performance throughout the life cycle of structural systems, as well as for supporting decision-making and maintenance planning. Although a great assortment of SHM methods has been developed, only a limited number of studies exist serving as reference basis for the comparison of different techniques. In this paper, the vibration-based assessment of a small-scale wind turbine (WT) blade is experimentally investigated, with the aim of establishing a benchmark case study for the SHM community. The structure under consideration, provided by Sonkyo Energy as part of the Windspot 3.5 kW WT model, is tested in both healthy and damaged states under varying environmental, that is, temperature, conditions as imposed by means of a climatic chamber. This study offers a thorough documentation of the configuration of this experimental benchmark, including the types of deployed sensors, the nature of excitation and available measurements, and the investigated damage scenarios and environmental variations enforced. Lastly, an overview of the raw and processed measurement data, made available to researchers via an open access Zenodo repository, is herein provided.

Stationary nonimaging lenses for solar concentration.

A novel approach for the design of refractive lenses is presented, where the lens is mounted on a stationary aperture and the Sun is tracked by a moving solar cell. The purpose of this work is to design a quasi-stationary concentrator by replacing the two-axis tracking of the Sun with internal motion of the miniaturized solar cell inside the module. Families of lenses are designed with a variation of the simultaneous multiple surface technique in which the sawtooth genetic algorithm is implemented to optimize the geometric variables of the optic in order to produce high fluxes for a range of incidence angles. Finally, we show examples of the technique for lenses with 60° and 30° acceptance half-angles, with low to medium attainable concentrations.

A modal derivatives enhanced Rubin substructuring method for geometrically nonlinear multibody systems.

This paper presents a novel model order reduction technique for 3D flexible multibody systems featuring nonlinear elastic behavior. We adopt the mean-axis floating frame approach in combination with an enhanced Rubin substructuring technique for the construction of the reduction basis. The standard Rubin reduction basis is augmented with the modal derivatives of both free-interface vibration modes and attachment modes to consider the bending-stretching coupling effects for each flexible body. The mean-axis frame generally yields relative displacements and rotations of smaller magnitude when compared to the one obtained by the nodal-fixed floating frame. This positively impacts the accuracy of the reduction basis. Also, when equipped with modal derivatives, the Rubin method better considers the geometric nonlinearities than the Craig-Bampton method, as it comprises vibration modes and modal derivatives featuring free motion of the interface. The nonlinear coupling between free-interface modes and attachment modes is also considered. Numerical tests confirm that the proposed method is more accurate than Craig-Bampton's, a nodal fixed floating frame counterpart originally proposed in Wu and Tiso (Multibody Syst. Dyn. 36(4): 405-425, [2016]), and produces significant speed-ups. However, the offline cost is increased because the mean-axis formulation produces operators with decreased sparsity patterns.

Behaviour of Prestressed CFRP Anchorages during and after Freeze-Thaw Cycle Exposure.

The long-term performance of externally-bonded reinforcements (EBR) on reinforced concrete (RC) structures highly depends on the behavior of constituent materials and their interfaces to various environmental loads, such as temperature and humidity exposure. Although significant efforts have been devoted to understanding the effect of such conditions on the anchorage resistance of unstressed EBR, with or without sustained loading, the effect of a released prestressing has not been thoroughly investigated. For this purpose, a series of experiments has been carried out herein, with concrete blocks strengthened with carbon fiber-reinforced polymer (CFRP) strips, both unstressed, as well as prestressed using the gradient anchorage. The gradient anchorage is a non-mechanical technique to anchor prestressed CFRP by exploiting the accelerated curing property of epoxy under higher temperatures and segment-wise prestress-force releasing. Subsequently, strengthened blocks are transferred into a chamber for exposure in dry freeze-thaw cycles (FTC). Following FTC exposure, the blocks are tested in a conventional lap-shear test setup to determine their residual anchorage resistance and then compared with reference specimens. Blocks were monitored during FTC by conventional and Fabry⁻Pérot-based fiber optic strain (FOS) sensors and a 3D-digital image correlation (3D-DIC) system during gradient application and lap-shear testing. Results indicate a reduction of residual anchorage resistance, stiffness and deformation capacity of the system after FTC and a change in the failure mode from concrete substrate to epoxy-concrete interface failure. It was further observed that all of these properties experienced a more significant reduction for prestressed specimens. These findings are presented with a complementary finite element model to shed more light onto the durability of such systems.

Multiple crack detection in 3D using a stable XFEM and global optimization.

A numerical scheme is proposed for the detection of multiple cracks in three dimensional (3D) structures. The scheme is based on a variant of the extended finite element method (XFEM) and a hybrid optimizer solution. The proposed XFEM variant is particularly well-suited for the simulation of 3D fracture problems, and as such serves as an efficient solution to the so-called forward problem. A set of heuristic optimization algorithms are recombined into a multiscale optimization scheme. The introduced approach proves effective in tackling the complex inverse problem involved, where identification of multiple flaws is sought on the basis of sparse measurements collected near the structural boundary. The potential of the scheme is demonstrated through a set of numerical case studies of varying complexity.