Fast prediction of propeller dynamic wake based on deep learning

Efficiently predicting the wake of propellers is of great importance for achieving propeller design optimization. In this work, the deep learning (DL) method called propeller wake convolutional neural networks (PWCNN) is proposed, which combines the transformer encoder and dilated convolutional bloc...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Physics of fluids (1994) 2024-08, Vol.36 (8)
Hauptverfasser: Li, Changming, Liang, Bingchen, Yuan, Peng, Zhang, Qin, Liu, Yongkai, Liu, Bin, Zhao, Ming
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue 8
container_start_page
container_title Physics of fluids (1994)
container_volume 36
creator Li, Changming
Liang, Bingchen
Yuan, Peng
Zhang, Qin
Liu, Yongkai
Liu, Bin
Zhao, Ming
description Efficiently predicting the wake of propellers is of great importance for achieving propeller design optimization. In this work, the deep learning (DL) method called propeller wake convolutional neural networks (PWCNN) is proposed, which combines the transformer encoder and dilated convolutional block to capture the multi-scale characteristics of wakes. Computational fluid dynamics (CFD) simulations are conducted using the delayed detached eddy simulation model for the wake to generate extensive high-fidelity wake data of the propeller operating under different operating conditions required for DL. PWCNN takes the wake predicted at the previous time step to update input and iteratively predicts the wake at future time steps to achieve dynamic wake prediction. The good agreement between DL prediction and CFD simulation results, with the mean relative error of the velocity components less than 2.36% for 15 future time steps, proves that PWCNN can efficiently capture the spatiotemporal evolution characteristic of dynamic wakes. Furthermore, PWCNN can predict the wake dynamic changes with reasonable accuracy under unseen operating conditions, further confirming the generality of the proposed model in forecasting the spatiotemporal evolution of propeller wake.
doi_str_mv 10.1063/5.0220551
format Article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_crossref_primary_10_1063_5_0220551</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>3089945611</sourcerecordid><originalsourceid>FETCH-LOGICAL-c182t-74bd87a727e45af9dddafe74670324a2bb3f700f46eebc0dee416f14f0663bf53</originalsourceid><addsrcrecordid>eNp9kE1LAzEQhoMoWKsH_0HAk8LWyXf3phSrQsGLnkN2M5Gt29012SL9901pz55mBh7m_SDklsGMgRaPagacg1LsjEwYzMvCaK3PD7uBQmvBLslVSmsAECXXE_K0dGmkQ0Tf1GPTd7QP-eoHbFuM1O86t2lq-ud-kFYuoacZ8YgDbdHFrum-r8lFcG3Cm9Ockq_ly-firVh9vL4vnldFzeZ8LIys_Nw4ww1K5ULpvXcBjdQGBJeOV5UIBiBIjVjVkCUk04HJANl0FZSYkrvj3-zud4tptOt-G7ssaUXOWUqlGcvU_ZGqY59SxGCH2Gxc3FkG9lCQVfZUUGYfjmyqm9Edwv8D7wHjfmSR</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3089945611</pqid></control><display><type>article</type><title>Fast prediction of propeller dynamic wake based on deep learning</title><source>AIP Journals Complete</source><creator>Li, Changming ; Liang, Bingchen ; Yuan, Peng ; Zhang, Qin ; Liu, Yongkai ; Liu, Bin ; Zhao, Ming</creator><creatorcontrib>Li, Changming ; Liang, Bingchen ; Yuan, Peng ; Zhang, Qin ; Liu, Yongkai ; Liu, Bin ; Zhao, Ming</creatorcontrib><description>Efficiently predicting the wake of propellers is of great importance for achieving propeller design optimization. In this work, the deep learning (DL) method called propeller wake convolutional neural networks (PWCNN) is proposed, which combines the transformer encoder and dilated convolutional block to capture the multi-scale characteristics of wakes. Computational fluid dynamics (CFD) simulations are conducted using the delayed detached eddy simulation model for the wake to generate extensive high-fidelity wake data of the propeller operating under different operating conditions required for DL. PWCNN takes the wake predicted at the previous time step to update input and iteratively predicts the wake at future time steps to achieve dynamic wake prediction. The good agreement between DL prediction and CFD simulation results, with the mean relative error of the velocity components less than 2.36% for 15 future time steps, proves that PWCNN can efficiently capture the spatiotemporal evolution characteristic of dynamic wakes. Furthermore, PWCNN can predict the wake dynamic changes with reasonable accuracy under unseen operating conditions, further confirming the generality of the proposed model in forecasting the spatiotemporal evolution of propeller wake.</description><identifier>ISSN: 1070-6631</identifier><identifier>EISSN: 1089-7666</identifier><identifier>DOI: 10.1063/5.0220551</identifier><identifier>CODEN: PHFLE6</identifier><language>eng</language><publisher>Melville: American Institute of Physics</publisher><subject>Artificial neural networks ; Computational fluid dynamics ; Deep learning ; Design optimization ; Detached eddy simulation ; Predictions ; Propellers ; Simulation models</subject><ispartof>Physics of fluids (1994), 2024-08, Vol.36 (8)</ispartof><rights>Author(s)</rights><rights>2024 Author(s). Published under an exclusive license by AIP Publishing.</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c182t-74bd87a727e45af9dddafe74670324a2bb3f700f46eebc0dee416f14f0663bf53</cites><orcidid>0000-0003-2971-8748 ; 0000-0002-8581-1042 ; 0000-0003-2000-2241 ; 0000-0003-1490-2243 ; 0000-0001-5010-9715</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,794,4512,27924,27925</link.rule.ids></links><search><creatorcontrib>Li, Changming</creatorcontrib><creatorcontrib>Liang, Bingchen</creatorcontrib><creatorcontrib>Yuan, Peng</creatorcontrib><creatorcontrib>Zhang, Qin</creatorcontrib><creatorcontrib>Liu, Yongkai</creatorcontrib><creatorcontrib>Liu, Bin</creatorcontrib><creatorcontrib>Zhao, Ming</creatorcontrib><title>Fast prediction of propeller dynamic wake based on deep learning</title><title>Physics of fluids (1994)</title><description>Efficiently predicting the wake of propellers is of great importance for achieving propeller design optimization. In this work, the deep learning (DL) method called propeller wake convolutional neural networks (PWCNN) is proposed, which combines the transformer encoder and dilated convolutional block to capture the multi-scale characteristics of wakes. Computational fluid dynamics (CFD) simulations are conducted using the delayed detached eddy simulation model for the wake to generate extensive high-fidelity wake data of the propeller operating under different operating conditions required for DL. PWCNN takes the wake predicted at the previous time step to update input and iteratively predicts the wake at future time steps to achieve dynamic wake prediction. The good agreement between DL prediction and CFD simulation results, with the mean relative error of the velocity components less than 2.36% for 15 future time steps, proves that PWCNN can efficiently capture the spatiotemporal evolution characteristic of dynamic wakes. Furthermore, PWCNN can predict the wake dynamic changes with reasonable accuracy under unseen operating conditions, further confirming the generality of the proposed model in forecasting the spatiotemporal evolution of propeller wake.</description><subject>Artificial neural networks</subject><subject>Computational fluid dynamics</subject><subject>Deep learning</subject><subject>Design optimization</subject><subject>Detached eddy simulation</subject><subject>Predictions</subject><subject>Propellers</subject><subject>Simulation models</subject><issn>1070-6631</issn><issn>1089-7666</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><recordid>eNp9kE1LAzEQhoMoWKsH_0HAk8LWyXf3phSrQsGLnkN2M5Gt29012SL9901pz55mBh7m_SDklsGMgRaPagacg1LsjEwYzMvCaK3PD7uBQmvBLslVSmsAECXXE_K0dGmkQ0Tf1GPTd7QP-eoHbFuM1O86t2lq-ud-kFYuoacZ8YgDbdHFrum-r8lFcG3Cm9Ockq_ly-firVh9vL4vnldFzeZ8LIys_Nw4ww1K5ULpvXcBjdQGBJeOV5UIBiBIjVjVkCUk04HJANl0FZSYkrvj3-zud4tptOt-G7ssaUXOWUqlGcvU_ZGqY59SxGCH2Gxc3FkG9lCQVfZUUGYfjmyqm9Edwv8D7wHjfmSR</recordid><startdate>202408</startdate><enddate>202408</enddate><creator>Li, Changming</creator><creator>Liang, Bingchen</creator><creator>Yuan, Peng</creator><creator>Zhang, Qin</creator><creator>Liu, Yongkai</creator><creator>Liu, Bin</creator><creator>Zhao, Ming</creator><general>American Institute of Physics</general><scope>AAYXX</scope><scope>CITATION</scope><scope>8FD</scope><scope>H8D</scope><scope>L7M</scope><orcidid>https://orcid.org/0000-0003-2971-8748</orcidid><orcidid>https://orcid.org/0000-0002-8581-1042</orcidid><orcidid>https://orcid.org/0000-0003-2000-2241</orcidid><orcidid>https://orcid.org/0000-0003-1490-2243</orcidid><orcidid>https://orcid.org/0000-0001-5010-9715</orcidid></search><sort><creationdate>202408</creationdate><title>Fast prediction of propeller dynamic wake based on deep learning</title><author>Li, Changming ; Liang, Bingchen ; Yuan, Peng ; Zhang, Qin ; Liu, Yongkai ; Liu, Bin ; Zhao, Ming</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c182t-74bd87a727e45af9dddafe74670324a2bb3f700f46eebc0dee416f14f0663bf53</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Artificial neural networks</topic><topic>Computational fluid dynamics</topic><topic>Deep learning</topic><topic>Design optimization</topic><topic>Detached eddy simulation</topic><topic>Predictions</topic><topic>Propellers</topic><topic>Simulation models</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Li, Changming</creatorcontrib><creatorcontrib>Liang, Bingchen</creatorcontrib><creatorcontrib>Yuan, Peng</creatorcontrib><creatorcontrib>Zhang, Qin</creatorcontrib><creatorcontrib>Liu, Yongkai</creatorcontrib><creatorcontrib>Liu, Bin</creatorcontrib><creatorcontrib>Zhao, Ming</creatorcontrib><collection>CrossRef</collection><collection>Technology Research Database</collection><collection>Aerospace Database</collection><collection>Advanced Technologies Database with Aerospace</collection><jtitle>Physics of fluids (1994)</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Li, Changming</au><au>Liang, Bingchen</au><au>Yuan, Peng</au><au>Zhang, Qin</au><au>Liu, Yongkai</au><au>Liu, Bin</au><au>Zhao, Ming</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Fast prediction of propeller dynamic wake based on deep learning</atitle><jtitle>Physics of fluids (1994)</jtitle><date>2024-08</date><risdate>2024</risdate><volume>36</volume><issue>8</issue><issn>1070-6631</issn><eissn>1089-7666</eissn><coden>PHFLE6</coden><abstract>Efficiently predicting the wake of propellers is of great importance for achieving propeller design optimization. In this work, the deep learning (DL) method called propeller wake convolutional neural networks (PWCNN) is proposed, which combines the transformer encoder and dilated convolutional block to capture the multi-scale characteristics of wakes. Computational fluid dynamics (CFD) simulations are conducted using the delayed detached eddy simulation model for the wake to generate extensive high-fidelity wake data of the propeller operating under different operating conditions required for DL. PWCNN takes the wake predicted at the previous time step to update input and iteratively predicts the wake at future time steps to achieve dynamic wake prediction. The good agreement between DL prediction and CFD simulation results, with the mean relative error of the velocity components less than 2.36% for 15 future time steps, proves that PWCNN can efficiently capture the spatiotemporal evolution characteristic of dynamic wakes. Furthermore, PWCNN can predict the wake dynamic changes with reasonable accuracy under unseen operating conditions, further confirming the generality of the proposed model in forecasting the spatiotemporal evolution of propeller wake.</abstract><cop>Melville</cop><pub>American Institute of Physics</pub><doi>10.1063/5.0220551</doi><tpages>16</tpages><orcidid>https://orcid.org/0000-0003-2971-8748</orcidid><orcidid>https://orcid.org/0000-0002-8581-1042</orcidid><orcidid>https://orcid.org/0000-0003-2000-2241</orcidid><orcidid>https://orcid.org/0000-0003-1490-2243</orcidid><orcidid>https://orcid.org/0000-0001-5010-9715</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 1070-6631
ispartof Physics of fluids (1994), 2024-08, Vol.36 (8)
issn 1070-6631
1089-7666
language eng
recordid cdi_crossref_primary_10_1063_5_0220551
source AIP Journals Complete
subjects Artificial neural networks
Computational fluid dynamics
Deep learning
Design optimization
Detached eddy simulation
Predictions
Propellers
Simulation models
title Fast prediction of propeller dynamic wake based on deep learning
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-04T12%3A22%3A05IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Fast%20prediction%20of%20propeller%20dynamic%20wake%20based%20on%20deep%20learning&rft.jtitle=Physics%20of%20fluids%20(1994)&rft.au=Li,%20Changming&rft.date=2024-08&rft.volume=36&rft.issue=8&rft.issn=1070-6631&rft.eissn=1089-7666&rft.coden=PHFLE6&rft_id=info:doi/10.1063/5.0220551&rft_dat=%3Cproquest_cross%3E3089945611%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=3089945611&rft_id=info:pmid/&rfr_iscdi=true