Pasar al contenido principal

HERRAMIENTAS DE INGENIERIA COMPUTACIONAL PARA PROSPECCION EN GEOTERMIA PROFUNDA

PID2023-153082OB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2023
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario CENTRE INTERNACIONAL DE METODES NUMERICS A L ENGINYERIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 8
Encontrada(s) 1 página(s)

UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Accelerating transient structural dynamics via SPILS-Net, a physics-derived latent space subdomain surrogate

  • Borst, Andino Raymond
  • Cavaliere, Faviola
  • Curtosi, Gabriel
  • Larráyoz Izcara, Xabier
  • Díez, Pedro|||0000-0001-6464-6407
  • Zlotnik, Sergio|||0000-0001-9674-8950
Complex, high-fidelity, transient numerical simulations often become computationally prohibitive for tasks requiring thousands of iterations, such as the rapid design space exploration necessitated by automotive crashworthiness assessments. Surrogate models have been widely adopted to overcome this limitation. However, traditional surrogate models and standard Recurrent Neural Networks (RNNs) struggle with complex history-dependent behaviors (e.g., elastoplasticity) due to uninterpretable latent states and the severe training costs of Backpropagation Through Time (BPTT). To address this, we propose the Spatiotemporal Physics-derived Internal Latent Space Network (SPILS-Net), a gray-box surrogate modeling framework. SPILS-Net explicitly models physical history using a physics-derived internal latent space, demonstrated herein via a PCA-reduced latent subspace of accumulated plastic strain. By employing a permanent “Teacher Forcing” strategy, the architecture shifts from sequence-based to increment-based learning. This completely bypasses BPTT and mitigates vanishing gradient problems, reconciling feed-forward training efficiency with the recursive depth required for stable, autonomous inference. Integrated into non-iterative Dirichlet–Neumann coupling workflows, SPILS-Net efficiently replaces expensive high-fidelity subdomains by directly predicting converged interface forces. Evaluated on a complex structural mechanics benchmark, the architecture achieved an 89% reduction in training time, dropping from 7.3 h for a baseline LSTM to 0.8 h, while simultaneously delivering superior predictive accuracy. By decoupling temporal sequence length from training cost, SPILS-Net provides a highly extensible, interpretable surrogate solution. Its architecture scales across extended time horizons, finer mesh resolutions, and increasingly complex physics, effectively unlocking the potential for extensive parameter studies in transient simulations., The authors would like to express their sincere gratitude to the anonymous reviewers for their thorough evaluation, insightful comments, and constructive feedback during the revision process, which significantly improved the quality of this final manuscript. The financial and technical support provided by SEAT S.A., facilitated through the International Centre for Numerical Methods in Engineering (CIMNE), is gratefully acknowledged. This work was partially supported by the Marie Sklodowska-Curie Actions (Doctoral Network with Grant agreement No. 101120556 to SZ and PD), the Spanish Ministry of Science, Innovation and Universities and the Spanish State Research Agency MICIU/AEI/10.13039/501100011033 (Grant agreement No. PID2023-153082OB-I00 to SZ and PD; PDC2025-165967-I00 to SZ, PD) and the Generalitat de Catalunya (Grant agreement No. 2021-SGR-01049)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Surrogates for physics-based and data-driven modelling of parametric systems: Review and new perspectives

  • Giacomini, Matteo|||0000-0001-6094-5944
  • Díez, Pedro|||0000-0001-6464-6407
Surrogate models provide compact relations between user-defined input parameters and output quantities of interest, enabling the efficient evaluation of complex parametric systems in many-query settings. Such capabilities are essential in a wide range of applications, including optimisation, control, data assimilation, uncertainty quantification, and emerging digital twin technologies in various fields such as manufacturing, personalised healthcare, smart cities, and sustainability. This article reviews established methodologies for constructing surrogate models exploiting either knowledge of the governing laws and the dynamical structure of the system (physics-based) or experimental observations (data-driven), as well as hybrid approaches combining these two paradigms. By revisiting the design of a surrogate model as a functional approximation problem, existing methodologies are reviewed in terms of the choice of (i) a reduced basis and (ii) a suitable approximation criterion. The paper reviews methodologies pertaining to the field of Scientific Machine Learning, and it aims at synthesising established knowledge, recent advances, and new perspectives on: dimensionality reduction, physics-based, and data-driven surrogate modelling based on proper orthogonal decomposition, proper generalised decomposition, and artificial neural networks; multi-fidelity methods to exploit information from sources with different fidelities; adaptive sampling, enrichment, and data augmentation techniques to enhance the quality of surrogate models., This work was partially supported by the Spanish Ministry of Science, Innovation and Universities and the Spanish State Research Agency MICIU/AEI/10.13039/501100011033 (Grant agreement No. TED2021-132021B-I00 to MG; PID2023-149979OB-I00 to MG; PID2023-153082OB-I00 to PD), EIT KIC RawMaterials Project Agreement 21123, European Commission MSCA Ref. 101,120,556-EARTHSAFE, and the Generalitat de Catalunya (Grant agreement No. 2021-SGR-01049). MG is Fellow of the Serra Hunter Programme of the Generalitat de Catalunya., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

A behavioral approach to direct data-driven fault detection

  • Markovsky, Ivan
  • Muixí Ballonga, Alba|||0000-0002-4420-3366
  • Zlotnik, Sergio|||0000-0001-9674-8950
  • Díez, Pedro|||0000-0001-6464-6407
© <2026> Elsevier. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/, Efficient and reliable fault detection methods are needed for monitoring and evaluation of processes, e.g., in structural health assessment. Existing methods rely on an a priori given model. Obtaining a model is nontrivial and is often the bottleneck in applications. Direct data-driven methods were recently developed in signal processing and control. These methods avoid parametric model identification and may outperform state-of-the-art model-based methods. This paper presents a direct data-driven method for fault detection. The monitored process is modeled as a linear time-invariant system with an unobserved deterministic disturbance. The behavioral approach to systems theory is used in order to define a representation invariant measure for the distance between data and model. The main contribution of the paper is a method for computing the distance directly from offline and online data without parametric model identification. Another contribution is a direct data-driven input estimation. The method is validated empirically on simulated data of a flexible beam with a crack, where the objective is to detect the crack., The research leading to these results received funding from: Catalan Institution for Research and Advanced Studies (ICREA), Spain, Fond for Scientific Research Vlaanderen (FWO), Spain project G033822N, Spanish Ministry of Science CIU/AEI/10.13039/501100011033/FEDER, UE, grants PID2023-148952OBI00, PID2023-153082OB-I00, and Generalitat de Catalunya, Spain, grant SGR 01049. AM is Fellow of the Serra Húnter Programme of the Generalitat de Catalunya.
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Adaptive reduced order modeling to assess peak stresses in heterogeneous arterial sections

  • Gahima, Stephan
  • Stefanati, Marco
  • Rodríguez Matas, José Félix
  • García González, Alberto|||0000-0003-2841-016X
  • Díez, Pedro|||0000-0001-6464-6407
We introduce an adaptive model reduction approach to compute peak Von Mises stress (pVMS) in heterogeneous arterial sections. The pipeline follows a standard two-phase process: first, we construct the training set of displacement snapshots obtained from the full order model offline, and then we compute pVMS in the online phase. We adaptively enrich the modal representation in critical regions (around the lumen and in calcified areas) using a level-set approach. Optimized for efficient pVMS computation, as a key plaque vulnerability indicator, this technique significantly reduces the computational cost of training machine learning models to classify plaque vulnerability based on pVMS., The authors acknowledge financial support provided by the Ministerio de Ciencia, Innovación y Universidades (MCIN/ AEI/10.13039/501100011033, MICIU/AEI/10.13039/501100011033/ FEDER, UE) through grants PID2022-141957OB-C21, PID2023-153082OB-I00 and CEX2018-000797-S; the Fondazione Regionale per la Ricerca Biomedica (Regione Lombardia), via project ID 3432721; and the Italian Ministry of University and Research (PRIN 2022, grant 2022ZKEP8S). The authors acknowledge Zhongzhao Teng of the University of Cambridge for providing the realistic 2D geometries used in this work. Funding Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature., Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

A behavioral approach to direct data-driven fault detection

  • Markovsky, Ivan
  • Muixí Ballonga, Alba
  • Zlotnik, Sergio
  • Díez, Pedro
© <2026> Elsevier. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/, Efficient and reliable fault detection methods are needed for monitoring and evaluation of processes, e.g., in structural health assessment. Existing methods rely on an a priori given model. Obtaining a model is nontrivial and is often the bottleneck in applications. Direct data-driven methods were recently developed in signal processing and control. These methods avoid parametric model identification and may outperform state-of-the-art model-based methods. This paper presents a direct data-driven method for fault detection. The monitored process is modeled as a linear time-invariant system with an unobserved deterministic disturbance. The behavioral approach to systems theory is used in order to define a representation invariant measure for the distance between data and model. The main contribution of the paper is a method for computing the distance directly from offline and online data without parametric model identification. Another contribution is a direct data-driven input estimation. The method is validated empirically on simulated data of a flexible beam with a crack, where the objective is to detect the crack., The research leading to these results received funding from: Catalan Institution for Research and Advanced Studies (ICREA), Spain, Fond for Scientific Research Vlaanderen (FWO), Spain project G033822N, Spanish Ministry of Science CIU/AEI/10.13039/501100011033/FEDER, UE, grants PID2023-148952OBI00, PID2023-153082OB-I00, and Generalitat de Catalunya, Spain, grant SGR 01049. AM is Fellow of the Serra Húnter Programme of the Generalitat de Catalunya., Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Accelerating transient structural dynamics via SPILS-Net, a physics-derived latent space subdomain surrogate

  • Borst, Andino Raymond
  • Cavaliere, Faviola
  • Curtosi, Gabriel
  • Larráyoz Izcara, Xabier
  • Díez, Pedro
  • Zlotnik, Sergio
Complex, high-fidelity, transient numerical simulations often become computationally prohibitive for tasks requiring thousands of iterations, such as the rapid design space exploration necessitated by automotive crashworthiness assessments. Surrogate models have been widely adopted to overcome this limitation. However, traditional surrogate models and standard Recurrent Neural Networks (RNNs) struggle with complex history-dependent behaviors (e.g., elastoplasticity) due to uninterpretable latent states and the severe training costs of Backpropagation Through Time (BPTT). To address this, we propose the Spatiotemporal Physics-derived Internal Latent Space Network (SPILS-Net), a gray-box surrogate modeling framework. SPILS-Net explicitly models physical history using a physics-derived internal latent space, demonstrated herein via a PCA-reduced latent subspace of accumulated plastic strain. By employing a permanent “Teacher Forcing” strategy, the architecture shifts from sequence-based to increment-based learning. This completely bypasses BPTT and mitigates vanishing gradient problems, reconciling feed-forward training efficiency with the recursive depth required for stable, autonomous inference. Integrated into non-iterative Dirichlet–Neumann coupling workflows, SPILS-Net efficiently replaces expensive high-fidelity subdomains by directly predicting converged interface forces. Evaluated on a complex structural mechanics benchmark, the architecture achieved an 89% reduction in training time, dropping from 7.3 h for a baseline LSTM to 0.8 h, while simultaneously delivering superior predictive accuracy. By decoupling temporal sequence length from training cost, SPILS-Net provides a highly extensible, interpretable surrogate solution. Its architecture scales across extended time horizons, finer mesh resolutions, and increasingly complex physics, effectively unlocking the potential for extensive parameter studies in transient simulations., The authors would like to express their sincere gratitude to the anonymous reviewers for their thorough evaluation, insightful comments, and constructive feedback during the revision process, which significantly improved the quality of this final manuscript. The financial and technical support provided by SEAT S.A., facilitated through the International Centre for Numerical Methods in Engineering (CIMNE), is gratefully acknowledged. This work was partially supported by the Marie Sklodowska-Curie Actions (Doctoral Network with Grant agreement No. 101120556 to SZ and PD), the Spanish Ministry of Science, Innovation and Universities and the Spanish State Research Agency MICIU/AEI/10.13039/501100011033 (Grant agreement No. PID2023-153082OB-I00 to SZ and PD; PDC2025-165967-I00 to SZ, PD) and the Generalitat de Catalunya (Grant agreement No. 2021-SGR-01049)., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Adaptive reduced order modeling to assess peak stresses in heterogeneous arterial sections

  • Gahima, Stephan
  • Stefanati, Marco
  • Rodríguez Matas, José Félix
  • García González, Alberto
  • Díez, Pedro
We introduce an adaptive model reduction approach to compute peak Von Mises stress (pVMS) in heterogeneous arterial sections. The pipeline follows a standard two-phase process: first, we construct the training set of displacement snapshots obtained from the full order model offline, and then we compute pVMS in the online phase. We adaptively enrich the modal representation in critical regions (around the lumen and in calcified areas) using a level-set approach. Optimized for efficient pVMS computation, as a key plaque vulnerability indicator, this technique significantly reduces the computational cost of training machine learning models to classify plaque vulnerability based on pVMS., The authors acknowledge financial support provided by the Ministerio de Ciencia, Innovación y Universidades (MCIN/ AEI/10.13039/501100011033, MICIU/AEI/10.13039/501100011033/ FEDER, UE) through grants PID2022-141957OB-C21, PID2023-153082OB-I00 and CEX2018-000797-S; the Fondazione Regionale per la Ricerca Biomedica (Regione Lombardia), via project ID 3432721; and the Italian Ministry of University and Research (PRIN 2022, grant 2022ZKEP8S). The authors acknowledge Zhongzhao Teng of the University of Cambridge for providing the realistic 2D geometries used in this work. Funding Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Surrogates for physics-based and data-driven modelling of parametric systems: Review and new perspectives

  • Giacomini, Matteo
  • Díez, Pedro
Surrogate models provide compact relations between user-defined input parameters and output quantities of interest, enabling the efficient evaluation of complex parametric systems in many-query settings. Such capabilities are essential in a wide range of applications, including optimisation, control, data assimilation, uncertainty quantification, and emerging digital twin technologies in various fields such as manufacturing, personalised healthcare, smart cities, and sustainability. This article reviews established methodologies for constructing surrogate models exploiting either knowledge of the governing laws and the dynamical structure of the system (physics-based) or experimental observations (data-driven), as well as hybrid approaches combining these two paradigms. By revisiting the design of a surrogate model as a functional approximation problem, existing methodologies are reviewed in terms of the choice of (i) a reduced basis and (ii) a suitable approximation criterion. The paper reviews methodologies pertaining to the field of Scientific Machine Learning, and it aims at synthesising established knowledge, recent advances, and new perspectives on: dimensionality reduction, physics-based, and data-driven surrogate modelling based on proper orthogonal decomposition, proper generalised decomposition, and artificial neural networks; multi-fidelity methods to exploit information from sources with different fidelities; adaptive sampling, enrichment, and data augmentation techniques to enhance the quality of surrogate models., This work was partially supported by the Spanish Ministry of Science, Innovation and Universities and the Spanish State Research Agency MICIU/AEI/10.13039/501100011033 (Grant agreement No. TED2021-132021B-I00 to MG; PID2023-149979OB-I00 to MG; PID2023-153082OB-I00 to PD), EIT KIC RawMaterials Project Agreement 21123, European Commission MSCA Ref. 101,120,556-EARTHSAFE, and the Generalitat de Catalunya (Grant agreement No. 2021-SGR-01049). MG is Fellow of the Serra Hunter Programme of the Generalitat de Catalunya., Peer Reviewed, Postprint (published version)