On Machine-Learning-Driven Surrogates for Sound Transmission Loss Simulations
Surrogate models are data-based approximations of computationally expensive simulations that enable efficient exploration of the model’s design space and informed decision making in many physical domains. The usage of surrogate models in the vibroacoustic domain, however, is challenging due to the n...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/12/21/10727 |
| _version_ | 1797469247120605184 |
|---|---|
| author | Barbara Zaparoli Cunha Abdel-Malek Zine Mohamed Ichchou Christophe Droz Stéphane Foulard |
| author_facet | Barbara Zaparoli Cunha Abdel-Malek Zine Mohamed Ichchou Christophe Droz Stéphane Foulard |
| author_sort | Barbara Zaparoli Cunha |
| collection | DOAJ |
| description | Surrogate models are data-based approximations of computationally expensive simulations that enable efficient exploration of the model’s design space and informed decision making in many physical domains. The usage of surrogate models in the vibroacoustic domain, however, is challenging due to the non-smooth, complex behavior of wave phenomena. This paper investigates four machine learning (ML) approaches in the modelling of surrogates of sound transmission loss (STL). Feature importance and feature engineering are used to improve the models’ accuracy while increasing their interpretability and physical consistency. The transfer of the proposed techniques to other problems in the vibroacoustic domain and possible limitations of the models are discussed. Experiments show that neural network surrogates with physics-guided features have better accuracy than other ML models across different STL models. Furthermore, sensitivity analysis methods are used to assess how physically coherent the analyzed surrogates are. |
| first_indexed | 2024-03-09T19:18:40Z |
| format | Article |
| id | doaj.art-00f0aafb4f6f48dc91efe50de44a20c5 |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-09T19:18:40Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-00f0aafb4f6f48dc91efe50de44a20c52023-11-24T03:31:51ZengMDPI AGApplied Sciences2076-34172022-10-0112211072710.3390/app122110727On Machine-Learning-Driven Surrogates for Sound Transmission Loss SimulationsBarbara Zaparoli Cunha0Abdel-Malek Zine1Mohamed Ichchou2Christophe Droz3Stéphane Foulard4Laboratory of Tribology and Dynamics of Systems, Ecole Centrale de Lyon, 69134 Ecully, FranceInstitut Camille Jordan, Ecole Centrale de Lyon, 69134 Ecully, FranceLaboratory of Tribology and Dynamics of Systems, Ecole Centrale de Lyon, 69134 Ecully, FranceCOSYS/SII, I4S Team, University Gustave Eiffel, Inria, 35042 Rennes, FranceCompredict GmbH, 64283 Darmstadt, GermanySurrogate models are data-based approximations of computationally expensive simulations that enable efficient exploration of the model’s design space and informed decision making in many physical domains. The usage of surrogate models in the vibroacoustic domain, however, is challenging due to the non-smooth, complex behavior of wave phenomena. This paper investigates four machine learning (ML) approaches in the modelling of surrogates of sound transmission loss (STL). Feature importance and feature engineering are used to improve the models’ accuracy while increasing their interpretability and physical consistency. The transfer of the proposed techniques to other problems in the vibroacoustic domain and possible limitations of the models are discussed. Experiments show that neural network surrogates with physics-guided features have better accuracy than other ML models across different STL models. Furthermore, sensitivity analysis methods are used to assess how physically coherent the analyzed surrogates are.https://www.mdpi.com/2076-3417/12/21/10727surrogatemachine learningsound transmission lossvibroacousticssensitivity analysisphysics-guided features |
| spellingShingle | Barbara Zaparoli Cunha Abdel-Malek Zine Mohamed Ichchou Christophe Droz Stéphane Foulard On Machine-Learning-Driven Surrogates for Sound Transmission Loss Simulations Applied Sciences surrogate machine learning sound transmission loss vibroacoustics sensitivity analysis physics-guided features |
| title | On Machine-Learning-Driven Surrogates for Sound Transmission Loss Simulations |
| title_full | On Machine-Learning-Driven Surrogates for Sound Transmission Loss Simulations |
| title_fullStr | On Machine-Learning-Driven Surrogates for Sound Transmission Loss Simulations |
| title_full_unstemmed | On Machine-Learning-Driven Surrogates for Sound Transmission Loss Simulations |
| title_short | On Machine-Learning-Driven Surrogates for Sound Transmission Loss Simulations |
| title_sort | on machine learning driven surrogates for sound transmission loss simulations |
| topic | surrogate machine learning sound transmission loss vibroacoustics sensitivity analysis physics-guided features |
| url | https://www.mdpi.com/2076-3417/12/21/10727 |
| work_keys_str_mv | AT barbarazaparolicunha onmachinelearningdrivensurrogatesforsoundtransmissionlosssimulations AT abdelmalekzine onmachinelearningdrivensurrogatesforsoundtransmissionlosssimulations AT mohamedichchou onmachinelearningdrivensurrogatesforsoundtransmissionlosssimulations AT christophedroz onmachinelearningdrivensurrogatesforsoundtransmissionlosssimulations AT stephanefoulard onmachinelearningdrivensurrogatesforsoundtransmissionlosssimulations |