Ponències/Comunicacions de congressos
http://hdl.handle.net/2117/6471
2024-09-19T13:49:20ZStructural health monitoring in a jacket-type wind turbine foundation: a minimum distortion embedding approach
http://hdl.handle.net/2117/412052
Structural health monitoring in a jacket-type wind turbine foundation: a minimum distortion embedding approach
Leon Medina, Jersson Xavier; Parés Mariné, Núria; Pozo Montero, Francesc
The extreme environmental conditions to which offshore wind turbine foundations are subjected make reliable monitoring methods necessary to predict possible structural damage. A data-driven approach was used to perform the structural health monitoring of a laboratory-scaled jacket-type wind turbine foundation. A white noise signal simulating the waves and wind in the sea was applied and amplified to the structure. The vibration-only response was measured by eight tri-axial accelerometers distributed in the structure. Five different structural classes, composed of the healthy and 4 unhealthy, were correctly classified using the following approach. 2460 measurements were acquired for the healthy structure and 820 by each one of the four unhealthy structures for 5740 measurements in total. The data was
arranged using an unfolded procedure resulting in a two-dimensional matrix. This resulting matrix has a size of 5740 x 58008. This resulting data has a high dimensionality. Therefore, using the minimum distortion embedding (MDE) approach, a dimensionality reduction procedure transforms the original data into a low dimensional space with fewer features. The low dimensional representation given by different distortion functions was compared changing the
repulsive and attractive penalties. The reduced feature matrix serves as input to a machine learning classifier. Several tree-based classifiers like decision trees, random forest and Adaboost were compared. A 5-fold cross validation procedure was executed to reduce the overfitting. Finally, classification accuracy was calculated as performance measure. The developed structural damage classification methodology yields high classification accuracies.
2024-07-18T12:34:26ZLeon Medina, Jersson XavierParés Mariné, NúriaPozo Montero, FrancescThe extreme environmental conditions to which offshore wind turbine foundations are subjected make reliable monitoring methods necessary to predict possible structural damage. A data-driven approach was used to perform the structural health monitoring of a laboratory-scaled jacket-type wind turbine foundation. A white noise signal simulating the waves and wind in the sea was applied and amplified to the structure. The vibration-only response was measured by eight tri-axial accelerometers distributed in the structure. Five different structural classes, composed of the healthy and 4 unhealthy, were correctly classified using the following approach. 2460 measurements were acquired for the healthy structure and 820 by each one of the four unhealthy structures for 5740 measurements in total. The data was
arranged using an unfolded procedure resulting in a two-dimensional matrix. This resulting matrix has a size of 5740 x 58008. This resulting data has a high dimensionality. Therefore, using the minimum distortion embedding (MDE) approach, a dimensionality reduction procedure transforms the original data into a low dimensional space with fewer features. The low dimensional representation given by different distortion functions was compared changing the
repulsive and attractive penalties. The reduced feature matrix serves as input to a machine learning classifier. Several tree-based classifiers like decision trees, random forest and Adaboost were compared. A 5-fold cross validation procedure was executed to reduce the overfitting. Finally, classification accuracy was calculated as performance measure. The developed structural damage classification methodology yields high classification accuracies.Numerical modelling of opposing subduction in the Western Mediterranean
http://hdl.handle.net/2117/408804
Numerical modelling of opposing subduction in the Western Mediterranean
Peral Millán, Mireia; Fernandez Ortiga, Manel; Vergés Masip, Jaume; Zlotnik, Sergio; Jimenez Munt, Ivone
The geodynamic evolution of the Western Mediterranean related to the closure of the LigurianTethys ocean is not yet fully resolved. We present a new 3D numerical model of double subduction with opposite polarities fostered by the inherited segmentation of the Ligurian-Tethys margins and rifting system between Iberia and NW Africa. The model is constrained by plate kinematic reconstructions and assumes that both Alboran-Tethys and Algerian-Tethys plate segments are separated by a NW-SE transform zone enabling that subduction polarity changes from SE-dipping in the Alboran-Tethys segment to NW-dipping in the Algerian-Tethys segment. The model starts about late Eocene times at 36.5 Ma and the temporal evolution of the simulation is tied to the geological evolution by comparing the rates of convergence and trench retreat, and the onset and end of opening in the Alboran Basin. Curvature of the Alboran-Tethys slab is imposed by the pinning of its western edge when reaching the end of the transform zone in the adjacent westAfrica continental block. The progressive curvature of the trench explains the observed regional stress reorientation changing from N-S to NW-SE and to E-W in the central and western regions of the Alboran Basin. The increase of the retreat rates from the Alboran-Tethys to the Algerian-Tethys slabs is compatible with the west-to-east transition from continental-to-magmatic-to-oceanic crustal nature and with the massive and partially synchronous calc-alkaline and alkaline magmatism.
2024-05-29T13:33:14ZPeral Millán, MireiaFernandez Ortiga, ManelVergés Masip, JaumeZlotnik, SergioJimenez Munt, IvoneThe geodynamic evolution of the Western Mediterranean related to the closure of the LigurianTethys ocean is not yet fully resolved. We present a new 3D numerical model of double subduction with opposite polarities fostered by the inherited segmentation of the Ligurian-Tethys margins and rifting system between Iberia and NW Africa. The model is constrained by plate kinematic reconstructions and assumes that both Alboran-Tethys and Algerian-Tethys plate segments are separated by a NW-SE transform zone enabling that subduction polarity changes from SE-dipping in the Alboran-Tethys segment to NW-dipping in the Algerian-Tethys segment. The model starts about late Eocene times at 36.5 Ma and the temporal evolution of the simulation is tied to the geological evolution by comparing the rates of convergence and trench retreat, and the onset and end of opening in the Alboran Basin. Curvature of the Alboran-Tethys slab is imposed by the pinning of its western edge when reaching the end of the transform zone in the adjacent westAfrica continental block. The progressive curvature of the trench explains the observed regional stress reorientation changing from N-S to NW-SE and to E-W in the central and western regions of the Alboran Basin. The increase of the retreat rates from the Alboran-Tethys to the Algerian-Tethys slabs is compatible with the west-to-east transition from continental-to-magmatic-to-oceanic crustal nature and with the massive and partially synchronous calc-alkaline and alkaline magmatism.Dynamic topography and satellite gravity data joint inversion using Reduced Order Models (DYGIRO)
http://hdl.handle.net/2117/408803
Dynamic topography and satellite gravity data joint inversion using Reduced Order Models (DYGIRO)
Ortega Gelabert, Olga; Fullea, Javier; Arnaiz Rodríguez, Mariano S.; Zlotnik, Sergio
Geophysical observables, such as surface elevation, gravity field anomalies, seismic data, surface heat flow, etc, are essential pieces of information used to make inferences about the structure and dynamics of the Earth’s interior. Simultaneously fitting different observable datasets is crucial in order to obtain consistent models. Among geophysical data, gravity data from ESA’s GOCE satellite mission provides key information in properly constraining the Earth’s density distribution. WINTERC-G is a new global thermochemical model of the lithosphere and upper mantle (currently being extended into the transition zone and lower mantle) based on terrestrial and satellite gravity data (Fullea et al., 2021). The inversion procedure behind WINTERC-G has two main steps. In step 1, a 1D column-wise inversion of surface wave tomographic, surface elevation (isostasy) and heat flow data is performed. Then, in step 2, the output model from step 1 is used as prior information for the inversion of the gravity field data (filtered geoid anomalies and gravity gradients from GOCE at satellite height) to refine the 3D crustal density and upper mantle composition. The model predicts a residual, non-isostatic topography that can be considered as a proxy for dynamic topography. However, within a rigorous framework, dynamic topography cannot be simply taken as a non- isostatic residual, but it should be explicitly computed (i.e. solving the Stokes equation for a given rheological and density distribution) and consistently integrated into the joint inversion of the gravity field and the terrestrial observation with feedback from both the static and dynamic parts. The goal of DYGIRO project is to add a third step into the global WINTERC-G inversion scheme that consistently integrates dynamic topography as an additional model constrain. We present here the first steps of such integration at global scale. To do that, the dynamic topography is computed by solving the Stokes flow problem associated with the current WINTERC-G model down to the transition zone. The dynamic topography thus obtained is coupled with the static thermochemical model constrained by gravity and seismic data within an iterative scheme where the observed surface elevation coincides with the model’s isostatic plus dynamic elevation contributions. The high computational cost associated with the large- scale 3D flow computations will be alleviated by means of Reduced Order Models. Such models are based on the idea of creating surrogate models that approximate the solution at a much lower computational cost.
2024-05-29T13:30:14ZOrtega Gelabert, OlgaFullea, JavierArnaiz Rodríguez, Mariano S.Zlotnik, SergioGeophysical observables, such as surface elevation, gravity field anomalies, seismic data, surface heat flow, etc, are essential pieces of information used to make inferences about the structure and dynamics of the Earth’s interior. Simultaneously fitting different observable datasets is crucial in order to obtain consistent models. Among geophysical data, gravity data from ESA’s GOCE satellite mission provides key information in properly constraining the Earth’s density distribution. WINTERC-G is a new global thermochemical model of the lithosphere and upper mantle (currently being extended into the transition zone and lower mantle) based on terrestrial and satellite gravity data (Fullea et al., 2021). The inversion procedure behind WINTERC-G has two main steps. In step 1, a 1D column-wise inversion of surface wave tomographic, surface elevation (isostasy) and heat flow data is performed. Then, in step 2, the output model from step 1 is used as prior information for the inversion of the gravity field data (filtered geoid anomalies and gravity gradients from GOCE at satellite height) to refine the 3D crustal density and upper mantle composition. The model predicts a residual, non-isostatic topography that can be considered as a proxy for dynamic topography. However, within a rigorous framework, dynamic topography cannot be simply taken as a non- isostatic residual, but it should be explicitly computed (i.e. solving the Stokes equation for a given rheological and density distribution) and consistently integrated into the joint inversion of the gravity field and the terrestrial observation with feedback from both the static and dynamic parts. The goal of DYGIRO project is to add a third step into the global WINTERC-G inversion scheme that consistently integrates dynamic topography as an additional model constrain. We present here the first steps of such integration at global scale. To do that, the dynamic topography is computed by solving the Stokes flow problem associated with the current WINTERC-G model down to the transition zone. The dynamic topography thus obtained is coupled with the static thermochemical model constrained by gravity and seismic data within an iterative scheme where the observed surface elevation coincides with the model’s isostatic plus dynamic elevation contributions. The high computational cost associated with the large- scale 3D flow computations will be alleviated by means of Reduced Order Models. Such models are based on the idea of creating surrogate models that approximate the solution at a much lower computational cost.Airlines network analysis on an air-rail multimodal system
http://hdl.handle.net/2117/407262
Airlines network analysis on an air-rail multimodal system
Delgado Muñoz, Luis; Gurtner, Gérald; Bolic, Tatjana; Trapote Barreira, César; Villardi de Montlaur, Adeline de
The network is a key asset for airlines because it defines their market power and constrains resource 12 allocation. In a hub and spoke network, short flights feed the airlines’ hubs to distribute passengers. 13 This network configuration is widely used by the industry as it increases their potential global con- 14 nectivity. Therefore, the analysis of the hub connectivity is paramount when assessing the efficiency 15 of long-haul airline’s schedules.
2024-04-29T13:39:43ZDelgado Muñoz, LuisGurtner, GéraldBolic, TatjanaTrapote Barreira, CésarVillardi de Montlaur, Adeline deThe network is a key asset for airlines because it defines their market power and constrains resource 12 allocation. In a hub and spoke network, short flights feed the airlines’ hubs to distribute passengers. 13 This network configuration is widely used by the industry as it increases their potential global con- 14 nectivity. Therefore, the analysis of the hub connectivity is paramount when assessing the efficiency 15 of long-haul airline’s schedules.OpenFOAM numerical analysis of the diurnal cycle of thermally-driven winds on Mars
http://hdl.handle.net/2117/399667
OpenFOAM numerical analysis of the diurnal cycle of thermally-driven winds on Mars
Arias Calderón, Santiago; Rojas Gregorio, José Ignacio; Villardi de Montlaur, Adeline de
The study of Martian winds holds significant scientific interest and potential practical applications. For instance, understanding the dynamics and behaviour of Martian winds is crucial in order to properly evaluate and select landing sites for missions to Mars [1]. The study of wind patterns is also critical in determining the capacity of Martian winds to remove dust from solar panels, thereby improving their performance and longevity [2]. Furthermore, Martian winds' effects on dust are fundamental to comprehending the planet's weather and climate [3], and finally, studying regional-scale sand transport and dune formation can contribute to gain a deeper understanding of Martian geological processes [4]. Therefore, the investigation of Martian winds represents a multidisciplinary approach with meaningful repercussions for planetary science and space exploration.
This work focuses on analysing the formation of thermally-driven winds, which are typically observed in inclined regions, examining their implications for the production of renewable wind energy. While wind energy may not be ideal for initial stages of human settlement on Mars, due mainly to the low density of the Martian atmosphere, it has the potential to serve as a renewable energy source in the long term and as a backup for solar energy [5]. Thermally-driven winds can attain considerably high velocities for steep slopes (values of up to 17 m/s [6]), exerting a dominant influence on the near-surface wind. The study of maximum velocities (Fig. 1) and the heights at which these velocities typically occur could be of particular interest for determining the optimal placement of wind energy resources (such as wind turbines) for future energy production on planet Mars.
Within the scope of this study, 2D simulations of slope winds on Mars are achieved using the open source computational fluid dynamics (CFD) code OpenFOAM. Several slope angles (ranging from 5º up to 20º) are considered. Following [7], the incompressible Navier-Stokes equations with Boussinesq approximation with a conventional k–e turbulence model are used. The impact of slope angle on the creation of thermal winds for both anabatic (up-slope) and katabatic (down-slope) flows is analysed: velocity and temperature profiles are presented for each slope angle along with the position of the maximum velocity. The numerical simulations shown in this work can serve as a cost-effective initial approximation, particularly when evaluating numerous configurations, while computationally more expensive techniques like Large Eddy Simulations (LES) or Direct Numerical Simulations (DNS) could be used in the future for validation and further exploration of final configurations.
2024-01-17T11:05:02ZArias Calderón, SantiagoRojas Gregorio, José IgnacioVillardi de Montlaur, Adeline deThe study of Martian winds holds significant scientific interest and potential practical applications. For instance, understanding the dynamics and behaviour of Martian winds is crucial in order to properly evaluate and select landing sites for missions to Mars [1]. The study of wind patterns is also critical in determining the capacity of Martian winds to remove dust from solar panels, thereby improving their performance and longevity [2]. Furthermore, Martian winds' effects on dust are fundamental to comprehending the planet's weather and climate [3], and finally, studying regional-scale sand transport and dune formation can contribute to gain a deeper understanding of Martian geological processes [4]. Therefore, the investigation of Martian winds represents a multidisciplinary approach with meaningful repercussions for planetary science and space exploration.
This work focuses on analysing the formation of thermally-driven winds, which are typically observed in inclined regions, examining their implications for the production of renewable wind energy. While wind energy may not be ideal for initial stages of human settlement on Mars, due mainly to the low density of the Martian atmosphere, it has the potential to serve as a renewable energy source in the long term and as a backup for solar energy [5]. Thermally-driven winds can attain considerably high velocities for steep slopes (values of up to 17 m/s [6]), exerting a dominant influence on the near-surface wind. The study of maximum velocities (Fig. 1) and the heights at which these velocities typically occur could be of particular interest for determining the optimal placement of wind energy resources (such as wind turbines) for future energy production on planet Mars.
Within the scope of this study, 2D simulations of slope winds on Mars are achieved using the open source computational fluid dynamics (CFD) code OpenFOAM. Several slope angles (ranging from 5º up to 20º) are considered. Following [7], the incompressible Navier-Stokes equations with Boussinesq approximation with a conventional k–e turbulence model are used. The impact of slope angle on the creation of thermal winds for both anabatic (up-slope) and katabatic (down-slope) flows is analysed: velocity and temperature profiles are presented for each slope angle along with the position of the maximum velocity. The numerical simulations shown in this work can serve as a cost-effective initial approximation, particularly when evaluating numerous configurations, while computationally more expensive techniques like Large Eddy Simulations (LES) or Direct Numerical Simulations (DNS) could be used in the future for validation and further exploration of final configurations.On the statement and numerical solution of the thermal problem within inversion methods for the study of lithospheric structure
http://hdl.handle.net/2117/394020
On the statement and numerical solution of the thermal problem within inversion methods for the study of lithospheric structure
Fernández, Mariano Tomás; Zlotnik, Sergio; Díez, Pedro
One of the goals of geophysicists is mapping and understanding the current structure of the Earth including its variations in composition, temperature and dynamical state. This structure is only accessible via indirect observations and, therefore, the mathematical problem to be solved is of an inverse kind. Within the inverse solver, many forward problems will be tested until finding a configuration compatible with the observations. This work deals with the problem statement and numerical solution of the forward thermal problem that arises from an inverse solver. In this case, we will use a simple parameterization of the Lithosphere-Asthenosphere Boundary (LAB), but the results are useful for other parametric description (e.g. one parameter per each cell).
A simplified model is used to show the ill-posedness of the mathematical problem arising when the LAB --an isotherm whose location is determined by the input parameters-- is imposed within the domain, over-constraining the forward problem. This is well-known in the community and several authors have proposed different approaches to circumvent it. Nevertheless, the strategies used in practice usually involve some non-physical procedures such as transitional regions where two different temperature fields are made compatible by smearing out differences. Generally, the solution in these regions does not comply with the governing equation and exhibits a non-physical behaviour.
In this work, we propose a specific problem statement for the temperature with interior essential conditions. The resulting problem is mathematically sound and results in a two-step numerical solver. This guarantees a self-consistent temperature field, in the sense that it respects the thermal governing equations everywhere.
The numerical domain is divided into two subdomains (lithosphere and asthenosphere) that are solved separately in the same mesh, using an unfitted mesh methodology. First, the temperature of the lithosphere is computed using the essential condition on the LAB. Second, the temperature in the mantle is obtained by minimizing a residual that measures the compatibility between the two subdomains in terms of LAB temperatures and across-LAB fluxes. This is done by adjusting the proper fluxes at the bottom of the numerical domain.
Several examples are presented showing that the obtained temperature fields are stable and oscillation-free. Moreover, the resulting fluxes at the bottom of the domain are reasonable and compatible with the expected values.
2023-09-26T07:48:32ZFernández, Mariano TomásZlotnik, SergioDíez, PedroOne of the goals of geophysicists is mapping and understanding the current structure of the Earth including its variations in composition, temperature and dynamical state. This structure is only accessible via indirect observations and, therefore, the mathematical problem to be solved is of an inverse kind. Within the inverse solver, many forward problems will be tested until finding a configuration compatible with the observations. This work deals with the problem statement and numerical solution of the forward thermal problem that arises from an inverse solver. In this case, we will use a simple parameterization of the Lithosphere-Asthenosphere Boundary (LAB), but the results are useful for other parametric description (e.g. one parameter per each cell).
A simplified model is used to show the ill-posedness of the mathematical problem arising when the LAB --an isotherm whose location is determined by the input parameters-- is imposed within the domain, over-constraining the forward problem. This is well-known in the community and several authors have proposed different approaches to circumvent it. Nevertheless, the strategies used in practice usually involve some non-physical procedures such as transitional regions where two different temperature fields are made compatible by smearing out differences. Generally, the solution in these regions does not comply with the governing equation and exhibits a non-physical behaviour.
In this work, we propose a specific problem statement for the temperature with interior essential conditions. The resulting problem is mathematically sound and results in a two-step numerical solver. This guarantees a self-consistent temperature field, in the sense that it respects the thermal governing equations everywhere.
The numerical domain is divided into two subdomains (lithosphere and asthenosphere) that are solved separately in the same mesh, using an unfitted mesh methodology. First, the temperature of the lithosphere is computed using the essential condition on the LAB. Second, the temperature in the mantle is obtained by minimizing a residual that measures the compatibility between the two subdomains in terms of LAB temperatures and across-LAB fluxes. This is done by adjusting the proper fluxes at the bottom of the numerical domain.
Several examples are presented showing that the obtained temperature fields are stable and oscillation-free. Moreover, the resulting fluxes at the bottom of the domain are reasonable and compatible with the expected values.Online structural damage classification methodology for offshore wind turbine foundations using data stream analysis
http://hdl.handle.net/2117/392625
Online structural damage classification methodology for offshore wind turbine foundations using data stream analysis
Leon Medina, Jersson Xavier; Parés Mariné, Núria; Pozo Montero, Francesc
Structural health monitoring (SHM) of wind turbines is crucial to improve maintenance and extend their lifespan. This study develops an online data analysis methodology using data stream analysis to classify damage in the links of an offshore wind turbine foundation. The methodology is validated using a laboratory-scaled jacket-type wind turbine foundation structure. 2460 measurements of the healthy structure were acquired, and a 5mm crack was applied to four different links to determine the four unhealthy classes. 820 measurements were taken for each of the unhealthy structures, resulting in a dataset with 5740 instances. As this is an imbalanced multiclass classification problem, a random sampler approach was used to treat the data. The only data obtained was from eight triaxial accelerometers distributed throughout the structure. Three different tree-based stream data classifiers were compared: Hoeffding Tree classifier, Extremely Fast Decision Tree classifier, and Hoeffding Adaptive Tree classifier. Each classification model underwent a tuning parameter procedure, and high values of the receiving operating characteristic area under the curve (ROC AUC) metric were achieved as a result. It is important to note that stream learning differs from batch learning.
2023-08-01T09:40:49ZLeon Medina, Jersson XavierParés Mariné, NúriaPozo Montero, FrancescStructural health monitoring (SHM) of wind turbines is crucial to improve maintenance and extend their lifespan. This study develops an online data analysis methodology using data stream analysis to classify damage in the links of an offshore wind turbine foundation. The methodology is validated using a laboratory-scaled jacket-type wind turbine foundation structure. 2460 measurements of the healthy structure were acquired, and a 5mm crack was applied to four different links to determine the four unhealthy classes. 820 measurements were taken for each of the unhealthy structures, resulting in a dataset with 5740 instances. As this is an imbalanced multiclass classification problem, a random sampler approach was used to treat the data. The only data obtained was from eight triaxial accelerometers distributed throughout the structure. Three different tree-based stream data classifiers were compared: Hoeffding Tree classifier, Extremely Fast Decision Tree classifier, and Hoeffding Adaptive Tree classifier. Each classification model underwent a tuning parameter procedure, and high values of the receiving operating characteristic area under the curve (ROC AUC) metric were achieved as a result. It is important to note that stream learning differs from batch learning.Ensemble of feature extraction methods to improve the structural damage classification in a wind turbine foundation
http://hdl.handle.net/2117/389282
Ensemble of feature extraction methods to improve the structural damage classification in a wind turbine foundation
Leon Medina, Jersson Xavier; Parés Mariné, Núria; Anaya, Maribel; Tibaduiza Burgos, Diego Alexander; Pozo Montero, Francesc
The condition monitoring of offshore wind power plants is an important topic that remains open. This monitoring search to lower the maintenance cost of this plants. One of the main components of the wind power plant is the wind turbine foundation. This study describes a data driven structural damage classification methodology applied in a wind turbine foundation. A vibration-response was captured in the structure using an accelerometer network. After arranged the obtained data a feature vector of 58,008 features was obtained. An ensemble approach of dimensionality reduction methods was applied to obtain a new set of features that is used to train a machine learning based classification model. Four different damage scenarios were applied in the structure. Therefore, considering the healthy structure, there are 5 classes in total that were correctly classified. As a result, 100\% of classification accuracy was obtained after applying the damage classification methodology in a wind-turbine offshore jacket-type foundation benchmark structure.
2023-06-21T07:08:00ZLeon Medina, Jersson XavierParés Mariné, NúriaAnaya, MaribelTibaduiza Burgos, Diego AlexanderPozo Montero, FrancescThe condition monitoring of offshore wind power plants is an important topic that remains open. This monitoring search to lower the maintenance cost of this plants. One of the main components of the wind power plant is the wind turbine foundation. This study describes a data driven structural damage classification methodology applied in a wind turbine foundation. A vibration-response was captured in the structure using an accelerometer network. After arranged the obtained data a feature vector of 58,008 features was obtained. An ensemble approach of dimensionality reduction methods was applied to obtain a new set of features that is used to train a machine learning based classification model. Four different damage scenarios were applied in the structure. Therefore, considering the healthy structure, there are 5 classes in total that were correctly classified. As a result, 100\% of classification accuracy was obtained after applying the damage classification methodology in a wind-turbine offshore jacket-type foundation benchmark structure.Imbalanced multi-class classification of structural damage in a wind turbine foundation
http://hdl.handle.net/2117/388551
Imbalanced multi-class classification of structural damage in a wind turbine foundation
Leon Medina, Jersson Xavier; Parés Mariné, Núria; Anaya, Maribel; Tibaduiza Burgos, Diego Alexander; Pozo Montero, Francesc
Damage diagnosis for offshore wind turbine foundations is a topic that remains open in the scientific community due to the importance of increasing safety and ensuring functionality. To deal with the challenge of online and in-service structural health monitoring (SHM) for wind turbines, approaches based on the vibration-response of the structure and captured by sensors such as the accelerometers need to be considered. This work presents a novel methodology to improve the structural damage classification of wind turbine foundations. This methodology consists of several stages. First, the data acquisition to collect and organize the information from the sensors attached to the structure, following the use of a mean-centered group scaling (MCGS) procedure to normalize the raw data and eliminate the difference between the magnitudes of the sensors. Next, a data unfolding to allow the multivariable analysis is performed. Then a linear feature extraction stage is applied to reduce the high dimensionality of the signals. Subsequently, the new feature array serves as the input to a supervised machine learning algorithm which allow to perform the classification task. A five-fold cross-validation procedure is used to obtain the goodness of classification. Several classification performance measures are calculated considering an imbalanced data set obtained with experimental data of a small-scale wind turbine foundation structure to validate the results of the proposed methodology.
2023-06-13T10:10:25ZLeon Medina, Jersson XavierParés Mariné, NúriaAnaya, MaribelTibaduiza Burgos, Diego AlexanderPozo Montero, FrancescDamage diagnosis for offshore wind turbine foundations is a topic that remains open in the scientific community due to the importance of increasing safety and ensuring functionality. To deal with the challenge of online and in-service structural health monitoring (SHM) for wind turbines, approaches based on the vibration-response of the structure and captured by sensors such as the accelerometers need to be considered. This work presents a novel methodology to improve the structural damage classification of wind turbine foundations. This methodology consists of several stages. First, the data acquisition to collect and organize the information from the sensors attached to the structure, following the use of a mean-centered group scaling (MCGS) procedure to normalize the raw data and eliminate the difference between the magnitudes of the sensors. Next, a data unfolding to allow the multivariable analysis is performed. Then a linear feature extraction stage is applied to reduce the high dimensionality of the signals. Subsequently, the new feature array serves as the input to a supervised machine learning algorithm which allow to perform the classification task. A five-fold cross-validation procedure is used to obtain the goodness of classification. Several classification performance measures are calculated considering an imbalanced data set obtained with experimental data of a small-scale wind turbine foundation structure to validate the results of the proposed methodology.Estimadores del error basados en estrellas
http://hdl.handle.net/2117/387311
Estimadores del error basados en estrellas
Parés Mariné, Núria
2023-05-11T13:13:41ZParés Mariné, Núria