Eleni Chatzi

• PhD
• Professor (Associate) at ETH Zurich

This work proposes an uncertainty-aware approach to the inverse problem of damage identification in a floating offshore wind turbine (FOWT). We design an autoencoder architecture, where the latent space represents the features of the target damage condition. The inverse operator (encoder) is a deep neural network that maps the measurable response t...

Modal identification is crucial for structural health monitoring and structural control, providing critical insights into structural dynamics and performance. This study presents a novel deep learning framework that integrates graph neural networks (GNNs), transformers, and a physics-informed loss function to achieve modal decomposition and identif...

The accurate modelling of structural dynamics is crucial across numerous engineering applications, such as Structural Health Monitoring (SHM), structural design optimisation, and vibration control. Often, these models originate from physics-based principles and can be derived from corresponding governing equations, often of differential equation fo...

The understanding and modeling of complex physical phenomena through dynamical systems has historically driven scientific progress, as it provides the tools for predicting the behavior of different systems under diverse conditions through time. The discovery of dynamical systems has been indispensable in engineering, as it allows for the analysis a...

Lamb waves offer a series of desirable features for SHM-applications, such as the ability to detect small defects, allowing to detect damage at early stages of its evolution. On the downside, their propagation through media with multiple geometrical features results in complicated patterns, which complicate the task of damage detection, thus hinder...

This work introduces a novel deep learning-based architecture, termed the Deep Belief Markov Model (DBMM), which provides efficient, model-formulation agnostic inference in Partially Observable Markov Decision Process (POMDP) problems. The POMDP framework allows for modeling and solving sequential decision-making problems under observation uncertai...

This study investigates the potential of using aerodynamic pressure time series measurements to detect structural damage in elastic, aerodynamically loaded structures. Our work is motivated by the increase in the dimensions of modern wind turbine blade designs, whose complex behavior necessitates the adoption of improved simulation and structural m...

In real-world scenarios, achieving domain adaptation and generalization poses significant challenges, as models must adapt to or generalize across unknown target distributions. Extending these capabilities to unseen multimodal distributions, i.e., multimodal domain adaptation and generalization, is even more challenging due to the distinct characte...

We introduce a Graph Transformer framework that serves as a general inverse physics engine on meshes, demonstrated through the challenging task of reconstructing aerodynamic flow fields from sparse surface measurements. While deep learning has shown promising results in forward physics simulation, inverse problems remain particularly challenging du...

Railway infrastructure is a crucial asset for the mobility of people and goods. The increased traffic frequency imposes higher loads and speeds, leading to accelerated infrastructure degradation. Asset managers require timely information regarding the current (diagnosis) and future (prognosis) condition of their assets to make informed decisions on...

Bonded iron-based shape memory alloy (Fe-SMA) strengthening has shown great potential in fatigue strengthening for steel structures. However, there is a lack of a comprehensive model for analysing the fatigue behaviour of systems strengthened with prestressed bonded Fe-SMA. This study proposes the first analytical model, integrating a prestress ana...

Test-time adaptation (TTA) has demonstrated significant potential in addressing distribution shifts between training and testing data. Open-set test-time adaptation (OSTTA) aims to adapt a source pre-trained model online to an unlabeled target domain that contains unknown classes. This task becomes more challenging when multiple modalities are invo...

This paper presents a data-driven model updating framework to estimate the operational parameters of a laterally-impacted pile. The goal is to facilitate the estimation of soil-pile interaction parameters such as the mobilized mass and stiffness, as well as geometrical data such as embedded pile length, using output-only information. Accurate knowl...

Floating Modular Energy Islands (FMEIs) are modularized, interconnected floating structures that function together to produce, store, possibly convert and transport renewable energy. Recent technological advancements in the offshore energy sector indicate that the concept of floating offshore energy islands has the potential to become more cost-eff...

This work proposes an uncertainty-aware approach to the inverse problem of damage identification in a Floating Offshore Wind Turbine (FOWT). We design an autoencoder architecture, where the latent space represents the features of the target damaged condition. The inverse operator (encoder) is a Deep Neural Network that maps the measurable response...

Anomaly detection (AD) is essential in identifying rare and often critical events in complex systems, finding applications in fields such as network intrusion detection, financial fraud detection, and fault detection in infrastructure and industrial systems. While AD is typically treated as an unsupervised learning task due to the high cost of labe...

In this paper, we provide a comprehensive definition and classification of various sources of uncertainty within the fields of Structural Dynamics, System Identification, and Structural Health Monitoring (SHM), with a primary focus on the latter. Utilizing the classical input-output system representation as a main contextual framework, we present a...

With global wind energy capacity ramping up, accurately predicting damage equivalent loads (DELs) and fatigue across wind turbine populations is critical, not only for ensuring the longevity of existing wind farms but also for the design of new farms. However, the estimation of such quantities of interests is hampered by the inherent complexity in...

The intersection of physics and machine learning has given rise to the physics-enhanced machine learning (PEML) paradigm, aiming to improve the capabilities and reduce the individual shortcomings of data- or physics-only methods. In this paper, the spectrum of PEML methods, expressed across the defining axes of physics and data, is discussed by eng...

Digital twins play an ever-increasing role in maximising the value of measurement and synthetic data by providing real-time monitoring of physical systems, integrating predictive models and creating actionable insights. This paper presents the development and implementation of the Aerosense digital twin for aerodynamic monitoring of wind turbine ro...

In this paper, we present a Bayesian framework for the identification of the parameters of nonlinear constitutive material laws using full-field displacement measurements. The concept of force-based Finite Element Model Updating (FEMU-F) is employed, which relies on the availability of measurable quantities such as displacements and external forces...

To maximize its value, the design, development and implementation of structural health monitoring (SHM) should focus on its role in facilitating decision support. In this position paper, we offer perspectives on the synergy between SHM and decision-making. We propose a classification of SHM use cases aligning with various dimensions that are closel...

Infrastructure systems are critical in modern communities but are highly susceptible to various natural and man-made disasters. Efficient post-disaster recovery requires repair-scheduling approaches under the limitation of capped resources that need to be shared across the system. Existing approaches, including component ranking methods, greedy evo...

The accurate modelling of structural dynamics is crucial across numerous engineering applications, such as Structural Health Monitoring (SHM), seismic analysis, and vibration control. Often, these models originate from physics-based principles and can be derived from corresponding governing equations, often of differential equation form. However, c...

Reinforced concrete structures featuring discontinuity regions are complex to design and often susceptible to errors linked to numerical analysis methods. For such structural design problems, strut-and-tie models offer a simple, intuitive and safe design method based on the lower bound theorem of plasticity. Although intuitive, the derivation of st...

Strut-and-tie models offer a simplified design approach for reinforced concrete structures such as walls or beams and are particularly suitable for static or geometrical discontinuities. They guarantee designs that are safe based on the lower bound theorem of the theory of plasticity. Currently, their manual generation demands significant time and...

Digital Twinning (DT) has become a main instrument for Industry 4.0 and the digital transformation of manufacturing and industrial processes. In this statement paper, we elaborate on the potential of Digital Twinning as a valuable tool in support of the management of intelligent infrastructures throughout all stages of their life-cycle. We highligh...

The rapid evolution of AI technologies over the past decades has profoundly impacted diverse sectors, including the engineering domain. AI techniques have already shown particular promise for handling geotechnical and underground engineering problems. These pose unique challenges stemming from the unpredictable nature of ground conditions, the intr...

Reduced Order Models (ROMs) form essential tools across engineering domains by virtue of their function as surrogates for computationally intensive digital twinning simulators. Although purely data-driven methods are available for ROM construction, schemes that allow to retain a portion of the physics tend to enhance the interpretability and genera...

The accuracy of the Kalman filter in state estimation depends on the knowledge of the process and measurement noise covariances. These are usually treated as tuning parameters and adjusted in a heuristic manner to fine-tune the state predictions. While several methods to identify the noise covariance from data exist, some require the use of optimiz...

The Population-Based Structural Health Monitoring (PBSHM) paradigm has recently emerged as a promising approach to enhance data-driven assessment of engineering structures by facilitating transfer learning between structures with some degree of similarity. In this work, we apply this concept to the automated modal identification of structural syste...

The task of open-set domain generalization (OSDG) involves recognizing novel classes within unseen domains, which becomes more challenging with multiple modalities as input. Existing works have only addressed unimodal OSDG within the meta-learning framework, without considering multimodal scenarios. In this work, we introduce a novel approach to ad...

A systematic multi-objective optimization approach is presented for designing vibration control systems for monopile Offshore Wind Turbines (OWT) under the combined actions of wind and wave. An extended configuration of the KDamper (EKD) is employed, with second-order tower phenomena and soil–structure interaction effects taken into account. A holi...

The performance of the Kalman filter is often hindered by the discrepancies between a model used to realize the filter and the true model of the data-generating system. While some methods to account for those errors exists, the majority is restricted to Luenberger's observers. The objective of this work is to develop a Kalman filter where the effec...

In the evolving landscape of sustainable energy, wind power has become a cornerstone in the transition towards renewable energy sources. With significant advancements in turbine technology and simulation methods, wind farms are now a cost-effective alternative to traditional power plants. Nevertheless, optimizing the performance and lifespan of win...

Reducing the Operation and Maintenance (O&M) costs of Wind Farms can substantially contribute towards adopting environmentally sustainable and cost effective energy sources. O&M costs in Wind Turbines can account up to 50% of the Turbine’s whole life cycle, with the majority of it being the frequent manual inspection needed for Wind Turbine Blades...

In the online structural health monitoring framework, surrogate modeling aims to approximate the predictions of the underlying high-fidelity model to facilitate fast simulations. Generally, conventional machine learning techniques based on purely data-driven approaches often behave as black boxes, producing predictions that may lack physical consis...

The health monitoring of railway networks offers a means to ensuring high-quality service, avoiding safety risks, and optimally planning maintenance actions to minimize life-cycle costs. Monitoring of the substructure is particularly linked to the tracking of degradation, which often stems from water infiltration, causing moisture accumulation in t...

Often, Structural Health Monitoring (SHM) campaigns draw damage-identifiers from vibration-based monitoring data. One of the downstream tasks of vibration-based SHM is that of system identification, i.e., inference of a model which is able to describe the system. For vibration-based monitoring, such models often rely on partial-differential equatio...

The global railway network spans over one million kilometers of tracks, and this extensive infrastructure is set to expand even further. The objective is to promote rail transportation as an environmentally sustainable solution to address the growing demands of mobility. A significant portion of these tracks spans across bridge structures, which mu...

The task of open-set domain generalization (OSDG) involves recognizing novel classes within unseen domains, which becomes more challenging with multiple modalities as input. Existing works have only addressed unimodal OSDG within the meta-learning framework, without considering multimodal scenarios. In this work, we introduce a novel approach to ad...

In this work, a novel approach of guided wave-based damage identification in composite laminates is proposed. The novelty of this research lies in the implementation of ConvLSTM-based autoencoders for the generation of full wavefield data of propagating guided waves in composite structures. The developed surrogate deep learning model takes as input...

In this paper, we present an unscented Kalman filter (UKF) for fusion of information from an accelerometer, global navigation satellite system (GNSS) instrumentation, and rotational sensor recordings of structural motion. Seismic and structural motions do not only include translations, but further incorporate torsion and twisting of the ground and/...

Condition monitoring of rotating machinery offers a salient tool for predictive maintenance on rolling elements, subjected to continuous working loads, wear, fatigue, and degradation. In this study, an enhanced computational tool for bearing fault simulation and feature extraction is proposed. A subsequent identification scheme is realized, through...

As offshore wind power expands globally, it is essential to ensure the reliable operation of components of such critical infrastructures. A less explored instance of such components, which are though essential in terms of operation, is found in subsea turbine cables and their protection systems, whose failure can incur prolonged shutdown periods an...

The rapid growth of the wind industry has resulted in larger wind turbines with modal properties that lie in the lower frequency range, rendering accurate fatigue assessment increasingly important. However, high uncertainty associated with the support conditions and foundation properties can pose challenges in the condition assessment and fatigue l...

Dynamic responses of steel bolts anchored in concrete and subjected to cyclic loads are complex: Qualitative physical attributes include pinching, displacement drift and force-intercept. In fact, these attributes vary with time, involving uncertainties that can lead to different failure patterns. This study explores higher-order elements (HOEs), an...

In this paper we focus on sensor placement for output-only modal analysis, where the objective is to choose those sensor locations yielding a minimal variance in the identification of modal parameters from measurement data. It is heuristically shown that the variance of modal parameters estimated with data-driven subspace identification can be appr...

Recently, researchers have proposed an innovative negative stiffness-based vibration control concept, namely the KDamper absorber. The envisaged mechanism comprises a combination of appropriate stiffness, damping and mass elements, including a negative stiffness element. Previous studies have formulated the mathematical framework of the system, as...

Understanding the dynamics of the interaction between railway vehicles and tracks is essential for forecasting vehicle and track conditions and performing maintenance actions to preserve the safety of railway infrastructure. In this work, physics-based models are deployed to predict the dynamic response of railway vehicles to track alignment and ir...

Modern wind turbines are large and slender dynamical structures with a fatigue loading profile of complex nature. The guarantee of their structural integrity is paramount for materializing cost efficient and more reliable wind energy. The measurement of the global dynamic response and loads of wind turbines is fundamental for achieving this goal. H...

Partially Observable Markov Decision Processes (POMDPs) can model complex sequential decision-making problems under stochastic and uncertain environments. A main reason hindering their broad adoption in real-world applications is the unavailability of a suitable POMDP model or a simulator thereof. Available solution algorithms, such as Reinforcemen...

The optimality of the Kalman filter for state-estimation depends on the knowledge of the process and measurement noise covariance. In applications, these covariances are often treated as tuning parameters, often adjusted in a heuristic manner based on user-defined performance criteria. While several methods to identify them from data exist, some re...

Detecting out-of-distribution (OOD) samples is important for deploying machine learning models in safety-critical applications such as autonomous driving and robot-assisted surgery. Existing research has mainly focused on unimodal scenarios on image data. However, real-world applications are inherently multimodal, which makes it essential to levera...

This study presents a robot-assisted solution for the automated identification of bridge frequencies and high-spatial-resolution mode shapes using a minimal number of sensors. The proposed approach employs programmable wheeled robots, whose movement can be remotely controlled, as the mobile platform carrying accelerometers. The output-only frequenc...

In this work, the propagation and attenuation of vertically polarized surface waves when interacting with terraced slopes is studied experimentally and numerically. To validate the devised simulation, a laboratory-scale physical model is tested in order to examine the attenuation properties of this well-known artificial landform. The experiment inv...

With the rapid evolution of the wind energy sector, there is an ever-increasing need to create value from the vast amounts of data made available both from within the domain and from other sectors. This article addresses the challenges faced by wind energy domain experts in converting data into domain knowledge, connecting and integrating them with...

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 subs...

Scientists from different disciplines at ETH Zurich are developing a dynamic, harmonised, and user-centred earthquake risk framework for Switzerland, relying on a continuously evolving earthquake catalogue generated by the Swiss Seismological Service (SED) using the national seismic networks. This framework uses all available information to assess...

In this paper, we incorporate the effect of nonlinear damping with the concept of locally resonant metamaterials to enable vibration attenuation beyond the conventional bandgap range. The proposed design combines a linear host cantilever beam and periodically distributed inertia amplifiers as nonlinear local resonators. The geometric nonlinearity i...

In this work, a novel approach is introduced for accelerating the solution of structural dynamics problems in the presence of localised phenomena, such as cracks. For this category of problems, conventional projection‐based Model Order Reduction (MOR) methods are either limited with respect to the range of system configurations that can be represen...

We propose coupling a physics-based reduction framework with a suited response decomposition technique to derive a component-oriented reduction (COR) approach, which is suitable for assembly systems featuring localized nonlinearities. Dependencies on influencing parameters are injected into the reduced-order model (ROM), thus ensuring robustness an...

Previous studies have demonstrated a great potential of prestressed strengthening of structures employing iron-based shape memory alloys (Fe-SMAs). A bonded Fe-SMA strengthening solution with partial activation has been proposed. However, an analytical model for assessing the strengthening efficiency was lacking, due to the unique nature of the emp...

To maximize its value, the design, development and implementation of Structural Health Monitoring (SHM) should focus on its role in facilitating decision support. In this position paper, we offer perspectives on the synergy between SHM and decision-making. We propose a classification of SHM use cases aligning with various dimensions that are closel...

With the rapid evolution of the wind energy sector, there is an ever-increasing need to create value from the vast amounts of data made available both from within the domain, as well as from other sectors. This article addresses the challenges faced by wind energy domain experts in converting data into domain knowledge, connecting and integrating i...