Programme Head
Claudio Sbarufatti
Associate Professor
Digital Twins for Health & Usage Monitoring and SHM
With the group
Shape sensing and damage detection of composite pressure vessels using inverse finite element method coupled with physics-based strain pre-extrapolation
J. Bardiani, R. Faure Ragani, L. Pinello, A. Kefal, A. Manes, C. Sbarufatti
Thin-Walled Structures, 218, 113935 · 2026
Structural damage diagnosis and prognosis with fleet digital twin considering similarity of individual structural features
J. Xu, D. Dai, X. Zhou, M. Giglio, C. Sbarufatti, L. Dong
Aerospace Science and Technology, 168, 110983 · 2026
Numerical model of curved composite tiles under low-velocity impact loading
S. Rezaei Akbarieh, D. Ma, C. Sbarufatti, A. Manes
Journal of Composite Materials · 2025
A hybrid approach to enhance decision-making in marine structures: Combining sensor data with human perception
J. Bardiani, C. Mazzolatti, A. Manes, C. Sbarufatti
Results in Engineering, 27, 105670 · 2025
Reference-free distributed monitoring of deflections in multi-span bridges
D. Poloni, M. Morgese, C. Wang, T. Taylor, M. Giglio, F. Ansari, C. Sbarufatti
Engineering Structures, 323, 119277 · 2025
Variational Neural Network Embedded with Digital Twins for Probabilistic Structural Damage Quantification
J. Xu, X. Zhou, M. Giglio, C. Sbarufatti, L. Dong
AIAA Journal, 63(6), 2474-2486 · 2025
On the effectiveness of ABH-based metamaterials in vibration control of naval equipment subjected to underwater explosion loads
J. Bardiani, G. Kyaw Oo D’Amore, G. Marchesi, M. Biot, C. Sbarufatti, A. Manes
Results in Engineering, 27, 106117 · 2025
On a meta-learning population-based approach to damage prognosis
G. Tsialiamanis, C. Sbarufatti, N. Dervilis, K. Worden
Mechanical Systems and Signal Processing, 209, 111119 · 2024
In-service Load Monitoring for an UAV Digital Twin
X. Zhou, M. Dziendzikowski, K. Dragan, L. Dong, M. Giglio, C. Sbarufatti
e-Journal of Nondestructive Testing, 29(7) · 2024
Preliminary Nose Landing Gear Digital Twin for Damage Detection
L. Pinello, O. Hassan, M. Giglio, C. Sbarufatti
Aerospace, 11(3), 222 · 2024
Towards a stochastic inverse Finite Element Method: A Gaussian Process strain extrapolation
D. Poloni, D. Oboe, C. Sbarufatti, M. Giglio
Mechanical Systems and Signal Processing, 189, 110056 · 2023
Variable Thickness Strain Pre-Extrapolation for the Inverse Finite Element Method
D. Poloni, D. Oboe, C. Sbarufatti, M. Giglio
Sensors, 23(3), 1733 · 2023
A fuzzy-set-based joint distribution adaptation method for regression and its application to online damage quantification for structural digital twin
X. Zhou, C. Sbarufatti, M. Giglio, L. Dong
Mechanical Systems and Signal Processing, 191, 110164 · 2023
Teaching
Lecturer
Digital Twin for Energy Systems Management
MSc Management of Built Environment; MSc Energy Engineering; MSc Management Engineering; MSc Industrial Safety and Risk Engineering · 8 CFU · 2nd semester · Milano Bovisa
Course sheetLecturer
Digital Twin for Health and Usage Monitoring
MSc Mechanical Engineering · 5 CFU · 1st semester · Milano Bovisa
Course sheetLecturer
Lab – Structural Health and Usage Monitoring in Action
MSc Mechanical Engineering · 5 CFU · 2nd semester · Milano Bovisa
Course sheetLecturer
Ship Structural Analysis and Design
MSc Mechanical Engineering · 10 CFU · 2nd semester · Milano Bovisa
Course sheet
Courses of 2026/2027, from the Politecnico’s course sheets. All the group’s courses
Open thesis topics
- SHMFleet MonitoringMachine LearningFatigue & FractureExperimentalSWARM
Population-Based SHM for Fatigue Damage Identification
This thesis focuses on the experimental validation of Population-Based Structural Health Monitoring (PBSHM) through fatigue crack propagation tests on a fleet of instrumented aluminum plates. The activity is designed from a fleet perspective, where multiple nominally identical specimens are tested under controlled conditions, with thickness as the only varying parameter to induce a structured domain shift. Each plate is equipped with a distributed network of strain sensors to monitor crack growth under cyclic loading. The objective is to develop and validate data-driven models capable of detecting damage, estimating structural health, and transferring knowledge across structurally related domains. Particular emphasis is placed on uncertainty quantification and cross-domain generalization. The work provides a controlled benchmark for assessing PBSHM capabilities before extending the methodology to more complex and realistic systems.
Details and apply - SHMFleet MonitoringMachine LearningExperimentalSWARM
Vibration-Based Fleet Monitoring under Structural and Environmental Variability
This thesis investigates vibration-based Population-Based Structural Health Monitoring (PBSHM) on a fleet of simplified aircraft-like structures. The systems share a common layout but differ in material, scale, or topology, introducing controlled heterogeneity. Each structure is instrumented with a limited number of accelerometers to reflect realistic sensing constraints. The experimental campaign includes dynamic testing, progressive damage introduction (mass addition, joint loosening), and systematic environmental variability (temperature and wind). The objective is to develop methods capable of distinguishing damage from operational and environmental effects, while enabling knowledge transfer across heterogeneous structures. The work provides a realistic intermediate validation step toward full-scale UAV applications.
Details and apply - SHMFleet MonitoringSensingUAVSWARM
Telemetry-Enhanced Fleet Monitoring for UAV Structural Health
This thesis investigates Population-Based Structural Health Monitoring (PBSHM) on a fleet of real quadrotor UAVs by combining onboard telemetry with additional sensors (e.g. accelerometers) integrated within payload constraints. The monitoring strategy leverages flight data together with selectively deployed sensing to enhance sensitivity to structural changes.
Details and apply - SHMFleet MonitoringMachine LearningWind Turbines
Population-Based Monitoring for Wind Turbine Structural Health
This thesis investigates Population-Based Structural Health Monitoring (PBSHM) applied to wind turbines using publicly available datasets from the literature. The activity focuses on fleets of turbines operating under varying environmental and operational conditions, exploiting signals such as vibration, SCADA, or load measurements.
Details and apply - SHMFleet MonitoringMachine LearningEngines
Population-Based SHM for Engine Fleet applications
Based on industrial data provided by the company on engine states and operations, the project aims to develop a Population-Based Structural Health Monitoring (PBSHM) system that leverages data from multiple similar engines to enhance monitoring, diagnostics, and maintenance efficiency. By sharing information across the engine population, the system will overcome data scarcity-enabling early fault detection even for engines with limited or no labeled data. In fact, some units provide rich datasets while others do not, yet all require consistent, reliable monitoring
Details and apply - Fleet MonitoringMachine LearningBatteriesSWARM
Domain Adaptation for transferring knowledge between Battery fleets
In a fleet of batteries, instead of treating each battery in isolation, PB-SHM exploits the shared behavior across a population of cells or modules. This enables learning from similarities and differences between batteries to improve diagnostic accuracy; by this way, PB-SHM allows knowledge gained from well-characterized batteries to be transferred to new or partially monitored batteries, using Domain Adaptation (DA) techniques like Transfer Component Analysis (TCA) or Joint Domain Adaptation (JDA). The same can be performed on multiple separated fleets of batteries (for example with different technologies), whether homogeneous or heterogeneous
Details and apply - Digital TwinsPHMAIReinforcement LearningUAVSWARM
Reinforcement Learning for Fleet O&M Optimization in UAV Systems
This thesis investigates the application of Reinforcement Learning (RL) to Operation and Maintenance (O&M) optimization in UAV fleets within a parametric simulation environment. The activity focuses on developing adaptive decision-making policies that optimise fleet-level performance under varying mission requirements and operational constraints.A flexible simulation framework is used to model different fleet configurations, enabling the training and evaluation of RL agents across multiple scenarios. An interactive interface allows direct comparison between RL-based decisions and user-driven strategies. The objective is to assess the capability of RL to improve key performance indicators such as fleet availability, mission efficiency, and robustness to changing conditions. Particular emphasis is placed on scalability, interpretability, and generalisation across configurations.
Details and apply - Digital TwinsAIReinforcement LearningBatteriesUAVSWARM
Reinforcement Learning for Battery-Aware Fleet Management in UAV Systems
This thesis investigates the application of Reinforcement Learning (RL) to the operational management of battery usage in UAV fleets. The activity focuses on learning decision policies that optimise fleet availability and mission continuity by managing the state-of-charge of individual UAVs.
Details and apply - Digital TwinsPHMEnginesTURBOMON
Development of an engine digital-twin
Optimising maintenance policies is a complex task pushing towards a shift from programmed maintenance to condition-based maintenance. An engine digital twin is thus suitable to monitor and keep track of the health condition throughout the lifespan of the engine, enabling the prediction of engine health state by implementing diagnostic and prognostic algorithms enabling conditions-based maintenance policies.
Details and apply - Digital TwinsPHMAIReinforcement LearningEnginesTURBOMON
Engine fleet logistic digital twin for life cycle management optimisation
Optimising maintenance policies is a complex task pushing towards a shift from programmed maintenance to condition-based maintenance. This task is fundamental when dealing with aircraft fleets, developing a Discrete Event Simulator (DES) to simulate numerous fleet lifecycles and implement Reinforcement Learning algorithms for automatic decision making.
Details and apply - Digital TwinsPHMAIReinforcement LearningWind Turbines
Reinforcement Learning for O&M Optimization in Wind Turbine Fleets
This thesis investigates the application of Reinforcement Learning (RL) to Operation and Maintenance (O&M) optimization in wind turbine fleets using publicly available datasets from the literature. The activity focuses on learning decision policies that optimise turbine availability and energy production under varying environmental and operational conditions.
Details and apply - SHMDigital TwinsiFEMBlastNumerical Modelling
Displacement Field Reconstruction Using the nonlinear iFEM Methodology under extreme loading conditions
Most structures operate in the linear-elastic range under normal service loads, where small-deformation assumptions hold. Under extreme loading (e.g., air-blast events), the response can become nonlinear (geometric and/or material), requiring methods that remain reliable beyond linear models. Building on strain data, this research aims to develop nonlinear iFEM techniques to reconstruct full-field displacements and identify damage under extreme loading conditions and geometric nonlinear cases.
Details and apply - SHMDigital TwinsiFEMSensing
Displacement Field Reconstruction under Torsional Loading Using iFEM-Based Sensor Networks
Torsional loading can occur in many engineering structures such as shafts, naval components, aerospace structures, and mechanical systems. Compared with bending, torsion produces complex strain distributions that are not easy to capture using conventional sparse sensing approaches. This thesis aims to investigate the design of efficient sensor networks for iFEM-based displacement field reconstruction under torsional and combined loading conditions (torsion + bending).
Details and apply - SHMDigital TwinsiFEM
Displacement Field Reconstruction of beam structures using iFEM considering degrading Boundary Conditions
In real structures, boundary conditions may degrade over time due to damage or deterioration, significantly affecting the structural response. This thesis focuses on the development of an iFEM-based framework for reconstructing the displacement field of beam structures considering degrading boundary conditions and support stiffness variations.
Details and apply - SHMDigital TwinsiFEMMachine Learning
Displacement Field Reconstruction Using the iFEM Methodology on a Rotating Shaft for Imbalance Detection
Building on strain data, this research aims to develop advanced iFEM methods for damage and imbalance identification by integrating physics-based models with machine learning techniques. The study will also incorporate pre-extrapolation methods to enhance prediction accuracy. Furthermore, adaptive strategies will be designed to accommodate varying shaft boundary conditions, ensuring robust and reliable full-field displacement reconstruction under realistic operational scenarios.
Details and apply - SHMDigital TwinsiFEMExperimentalNaval
iFEM-Based Battle Damage Identification in Naval Structures
This thesis investigates the application of the inverse Finite Element Method (iFEM) for real-time damage identification in naval structures using experimental strain data. The activity is based on measurements acquired from a scaled structural model tested in a controlled wave tank under different sea states and damage scenarios. The objective is to reconstruct the structural response from strain measurements and detect damage conditions through deviations in the estimated displacement and stress fields. Different levels of structural damage and environmental loading are considered. Particular emphasis is placed on robustness to operational variability, sensitivity to damage, and the integration of iFEM within data-driven monitoring frameworks.
Details and apply - SHMPHMMachine LearningHelicoptersSAMAS 2
Data-based monitoring for impact/corrosion damage identification in helicopters
In the framework of SAMAS 2, flight tests with real helicopters are performed to evaluate the health status through a specifically designed Structural Health Monitoring and Prognosis system. The data from the flight tests are used to develop diagnostic algorithms capable of identifying damage and the degradation level of the structure.
Details and apply - SHMSensingImpact & BallisticsCompositesExperimental
Self-Sensing Composite Structures for Impact Damage Detection Using CNT Buckypapers
This thesis investigates impact damage detection, localisation, and characterisation in heterogeneous composite structures through the integration of self-sensing layers. A composite specimen is designed and manufactured using multiple material phases (e.g. CFRP and ceramic-like layers), embedding CNT buckypapers at different depths within the laminate.
Details and apply - Quantum ComputingNumerical Modelling
Development of quantum algorithms for Engineering (FEM)
Quantum computing is a novel, promising technology that may have a great impact also in the Engineering field. Nowadays, it is in its first development stages, while the potential and applicability of such technology still have to be determined. This thesis project aims to study the latest innovations in terms of quantum computing and reproduce common traditional algorithms with quantum ones. The goal is to reproduce an FEM algorithm in a simple scenario and evaluate possible extensions to more complex ones, considering the actual technological limitations.
Details and apply - AIMachine LearningQuantum Computing
Development of quantum machine learning (QML) for diagnostics
Quantum computing is a novel, promising technology that may have a great impact also in the Engineering field. Nowadays, it is in its first development stages, while the potential and applicability of such technology still have to be determined. This thesis project aims to study the latest innovations in terms of quantum computing and reproduce common traditional algorithms with quantum ones. The goal is to develop a Quantum Machine Learning (QML) algorithm to perform damage detection on dataset already available in our lab.
Details and apply