Recurrent neural network model with Bayesian training and mutual information for response prediction of large buildings
•A new methodology for accurate response prediction of large structures is proposed.•It uses EMD, MI index, and a probabilistic Bayesian-based training algorithm.•An MI index is proposed to determine the optimum number of neurons in the NN model.•Bayesian regularization is proposed to train the opti...
Gespeichert in:
| Veröffentlicht in: | Engineering structures 2019-01, Vol.178, p.603-615 |
|---|---|
| Hauptverfasser: | , , , , , |
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 615 |
|---|---|
| container_issue | |
| container_start_page | 603 |
| container_title | Engineering structures |
| container_volume | 178 |
| creator | Perez-Ramirez, Carlos A. Amezquita-Sanchez, Juan P. Valtierra-Rodriguez, Martin Adeli, Hojjat Dominguez-Gonzalez, Aurelio Romero-Troncoso, Rene J. |
| description | •A new methodology for accurate response prediction of large structures is proposed.•It uses EMD, MI index, and a probabilistic Bayesian-based training algorithm.•An MI index is proposed to determine the optimum number of neurons in the NN model.•Bayesian regularization is proposed to train the optimized NN model.•It is applied to a 1:20-scaled 38-story highrise building and a 5-story steel frame.
An accurate response prediction model is of great importance in various applications such as damage detection, structural health monitoring, and vibration control. Development of such a methodology for large civil structures is challenging because of their size and complicated behavior and noise-contaminated, nonlinear, and nonstationary nature of the signals. In addition, the prediction model must have a low computational burden for real-time applications. In this article, a new methodology and a nonlinear autoregressive exogenous model (NARX)-based recurrent neural network (NN) model is presented for accurate response prediction of large structures. The methodology is based on adroit integration of three concepts: a recent signal processing concept, empirical mode decomposition (EMD), mutual information (MI) index from the information theory, and a probabilistic Bayesian-based training algorithm. The EMD method is used to remove the noise in the measured signals. An MI index is proposed to determine the optimum number of neurons in the hidden layer of the NN model with the goal of reducing the computational requirements without affecting its performance. Finally, Bayesian regularization (BR) is proposed to train the optimized NN model. The effectiveness of the proposed methodology is assessed by predicting the structural response of a 1:20-scaled 38-story highrise building structure subjected to seismic excitations and ambient vibrations, and a five-story steel frame subjected to different levels of the Kobe earthquake. |
| doi_str_mv | 10.1016/j.engstruct.2018.10.065 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_journals_2166740242</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0141029618307235</els_id><sourcerecordid>2166740242</sourcerecordid><originalsourceid>FETCH-LOGICAL-c343t-35c73c82f5861c7f90ce6edaba667c6aa6380f948e30e428e94fd278e511500a3</originalsourceid><addsrcrecordid>eNqFUNFOGzEQtFArkUK_oZZ4vnRt3_mcR0AtRUJCQvBsGd9e6vRih7WvEX9fhyBe-zSr3ZlZzTD2TcBSgNDfN0uM61xo9mUpQZi6XYLuTthCmF41vZLqE1uAaEUDcqVP2ZecNwAgjYEF2z-gn4kwFh5xJjdVKPtEf_g2DTjxfSi_-ZV7xRxc5IVciCGuuYsD385lrvwQx0RbV0KKvE6cMO9SzMh3hEPwb_s08snRGvnzHKahGuRz9nl0U8av73jGnn7-eLz-1dzd39xeX941XrWqNKrzvfJGjp3RwvfjCjxqHNyz07r32jmtDIyr1qACbKXBVTsOsjfYCdEBOHXGLo6-O0ovM-ZiN2mmWF9aKapHC7KVldUfWZ5SzoSj3VHYOnq1AuyhZbuxHy3bQ8uHQ225Ki-PSqwh_gYkm33A6Gt0wsodUvivxz_7do1K</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2166740242</pqid></control><display><type>article</type><title>Recurrent neural network model with Bayesian training and mutual information for response prediction of large buildings</title><source>ScienceDirect Journals (5 years ago - present)</source><creator>Perez-Ramirez, Carlos A. ; Amezquita-Sanchez, Juan P. ; Valtierra-Rodriguez, Martin ; Adeli, Hojjat ; Dominguez-Gonzalez, Aurelio ; Romero-Troncoso, Rene J.</creator><creatorcontrib>Perez-Ramirez, Carlos A. ; Amezquita-Sanchez, Juan P. ; Valtierra-Rodriguez, Martin ; Adeli, Hojjat ; Dominguez-Gonzalez, Aurelio ; Romero-Troncoso, Rene J.</creatorcontrib><description>•A new methodology for accurate response prediction of large structures is proposed.•It uses EMD, MI index, and a probabilistic Bayesian-based training algorithm.•An MI index is proposed to determine the optimum number of neurons in the NN model.•Bayesian regularization is proposed to train the optimized NN model.•It is applied to a 1:20-scaled 38-story highrise building and a 5-story steel frame.
An accurate response prediction model is of great importance in various applications such as damage detection, structural health monitoring, and vibration control. Development of such a methodology for large civil structures is challenging because of their size and complicated behavior and noise-contaminated, nonlinear, and nonstationary nature of the signals. In addition, the prediction model must have a low computational burden for real-time applications. In this article, a new methodology and a nonlinear autoregressive exogenous model (NARX)-based recurrent neural network (NN) model is presented for accurate response prediction of large structures. The methodology is based on adroit integration of three concepts: a recent signal processing concept, empirical mode decomposition (EMD), mutual information (MI) index from the information theory, and a probabilistic Bayesian-based training algorithm. The EMD method is used to remove the noise in the measured signals. An MI index is proposed to determine the optimum number of neurons in the hidden layer of the NN model with the goal of reducing the computational requirements without affecting its performance. Finally, Bayesian regularization (BR) is proposed to train the optimized NN model. The effectiveness of the proposed methodology is assessed by predicting the structural response of a 1:20-scaled 38-story highrise building structure subjected to seismic excitations and ambient vibrations, and a five-story steel frame subjected to different levels of the Kobe earthquake.</description><identifier>ISSN: 0141-0296</identifier><identifier>EISSN: 1873-7323</identifier><identifier>DOI: 10.1016/j.engstruct.2018.10.065</identifier><language>eng</language><publisher>Kidlington: Elsevier Ltd</publisher><subject>Aseismic buildings ; Autoregressive models ; Bayesian analysis ; Computation ; Computational mathematics ; Damage detection ; Data processing ; Earthquake damage ; Earthquakes ; Empirical mode decomposition ; High rise buildings ; Highrise building structures ; Information processing ; Information theory ; Methodology ; Neural networks ; Non-linear autoregressive exogenous model ; Prediction models ; Recurrent neural networks ; Regularization ; Seismic activity ; Seismic engineering ; Signal processing ; Steel frames ; Steel structures ; Structural damage ; Structural health monitoring ; Structural system identification ; Vibration analysis ; Vibration control ; Vibration monitoring ; Vibrations</subject><ispartof>Engineering structures, 2019-01, Vol.178, p.603-615</ispartof><rights>2018 Elsevier Ltd</rights><rights>Copyright Elsevier BV Jan 1, 2019</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c343t-35c73c82f5861c7f90ce6edaba667c6aa6380f948e30e428e94fd278e511500a3</citedby><cites>FETCH-LOGICAL-c343t-35c73c82f5861c7f90ce6edaba667c6aa6380f948e30e428e94fd278e511500a3</cites><orcidid>0000-0003-3839-1396 ; 0000-0001-5718-1453</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://dx.doi.org/10.1016/j.engstruct.2018.10.065$$EHTML$$P50$$Gelsevier$$H</linktohtml><link.rule.ids>314,780,784,3550,27924,27925,45995</link.rule.ids></links><search><creatorcontrib>Perez-Ramirez, Carlos A.</creatorcontrib><creatorcontrib>Amezquita-Sanchez, Juan P.</creatorcontrib><creatorcontrib>Valtierra-Rodriguez, Martin</creatorcontrib><creatorcontrib>Adeli, Hojjat</creatorcontrib><creatorcontrib>Dominguez-Gonzalez, Aurelio</creatorcontrib><creatorcontrib>Romero-Troncoso, Rene J.</creatorcontrib><title>Recurrent neural network model with Bayesian training and mutual information for response prediction of large buildings</title><title>Engineering structures</title><description>•A new methodology for accurate response prediction of large structures is proposed.•It uses EMD, MI index, and a probabilistic Bayesian-based training algorithm.•An MI index is proposed to determine the optimum number of neurons in the NN model.•Bayesian regularization is proposed to train the optimized NN model.•It is applied to a 1:20-scaled 38-story highrise building and a 5-story steel frame.
An accurate response prediction model is of great importance in various applications such as damage detection, structural health monitoring, and vibration control. Development of such a methodology for large civil structures is challenging because of their size and complicated behavior and noise-contaminated, nonlinear, and nonstationary nature of the signals. In addition, the prediction model must have a low computational burden for real-time applications. In this article, a new methodology and a nonlinear autoregressive exogenous model (NARX)-based recurrent neural network (NN) model is presented for accurate response prediction of large structures. The methodology is based on adroit integration of three concepts: a recent signal processing concept, empirical mode decomposition (EMD), mutual information (MI) index from the information theory, and a probabilistic Bayesian-based training algorithm. The EMD method is used to remove the noise in the measured signals. An MI index is proposed to determine the optimum number of neurons in the hidden layer of the NN model with the goal of reducing the computational requirements without affecting its performance. Finally, Bayesian regularization (BR) is proposed to train the optimized NN model. The effectiveness of the proposed methodology is assessed by predicting the structural response of a 1:20-scaled 38-story highrise building structure subjected to seismic excitations and ambient vibrations, and a five-story steel frame subjected to different levels of the Kobe earthquake.</description><subject>Aseismic buildings</subject><subject>Autoregressive models</subject><subject>Bayesian analysis</subject><subject>Computation</subject><subject>Computational mathematics</subject><subject>Damage detection</subject><subject>Data processing</subject><subject>Earthquake damage</subject><subject>Earthquakes</subject><subject>Empirical mode decomposition</subject><subject>High rise buildings</subject><subject>Highrise building structures</subject><subject>Information processing</subject><subject>Information theory</subject><subject>Methodology</subject><subject>Neural networks</subject><subject>Non-linear autoregressive exogenous model</subject><subject>Prediction models</subject><subject>Recurrent neural networks</subject><subject>Regularization</subject><subject>Seismic activity</subject><subject>Seismic engineering</subject><subject>Signal processing</subject><subject>Steel frames</subject><subject>Steel structures</subject><subject>Structural damage</subject><subject>Structural health monitoring</subject><subject>Structural system identification</subject><subject>Vibration analysis</subject><subject>Vibration control</subject><subject>Vibration monitoring</subject><subject>Vibrations</subject><issn>0141-0296</issn><issn>1873-7323</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><recordid>eNqFUNFOGzEQtFArkUK_oZZ4vnRt3_mcR0AtRUJCQvBsGd9e6vRih7WvEX9fhyBe-zSr3ZlZzTD2TcBSgNDfN0uM61xo9mUpQZi6XYLuTthCmF41vZLqE1uAaEUDcqVP2ZecNwAgjYEF2z-gn4kwFh5xJjdVKPtEf_g2DTjxfSi_-ZV7xRxc5IVciCGuuYsD385lrvwQx0RbV0KKvE6cMO9SzMh3hEPwb_s08snRGvnzHKahGuRz9nl0U8av73jGnn7-eLz-1dzd39xeX941XrWqNKrzvfJGjp3RwvfjCjxqHNyz07r32jmtDIyr1qACbKXBVTsOsjfYCdEBOHXGLo6-O0ovM-ZiN2mmWF9aKapHC7KVldUfWZ5SzoSj3VHYOnq1AuyhZbuxHy3bQ8uHQ225Ki-PSqwh_gYkm33A6Gt0wsodUvivxz_7do1K</recordid><startdate>20190101</startdate><enddate>20190101</enddate><creator>Perez-Ramirez, Carlos A.</creator><creator>Amezquita-Sanchez, Juan P.</creator><creator>Valtierra-Rodriguez, Martin</creator><creator>Adeli, Hojjat</creator><creator>Dominguez-Gonzalez, Aurelio</creator><creator>Romero-Troncoso, Rene J.</creator><general>Elsevier Ltd</general><general>Elsevier BV</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7SR</scope><scope>7ST</scope><scope>8BQ</scope><scope>8FD</scope><scope>C1K</scope><scope>FR3</scope><scope>JG9</scope><scope>KR7</scope><scope>SOI</scope><orcidid>https://orcid.org/0000-0003-3839-1396</orcidid><orcidid>https://orcid.org/0000-0001-5718-1453</orcidid></search><sort><creationdate>20190101</creationdate><title>Recurrent neural network model with Bayesian training and mutual information for response prediction of large buildings</title><author>Perez-Ramirez, Carlos A. ; Amezquita-Sanchez, Juan P. ; Valtierra-Rodriguez, Martin ; Adeli, Hojjat ; Dominguez-Gonzalez, Aurelio ; Romero-Troncoso, Rene J.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c343t-35c73c82f5861c7f90ce6edaba667c6aa6380f948e30e428e94fd278e511500a3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Aseismic buildings</topic><topic>Autoregressive models</topic><topic>Bayesian analysis</topic><topic>Computation</topic><topic>Computational mathematics</topic><topic>Damage detection</topic><topic>Data processing</topic><topic>Earthquake damage</topic><topic>Earthquakes</topic><topic>Empirical mode decomposition</topic><topic>High rise buildings</topic><topic>Highrise building structures</topic><topic>Information processing</topic><topic>Information theory</topic><topic>Methodology</topic><topic>Neural networks</topic><topic>Non-linear autoregressive exogenous model</topic><topic>Prediction models</topic><topic>Recurrent neural networks</topic><topic>Regularization</topic><topic>Seismic activity</topic><topic>Seismic engineering</topic><topic>Signal processing</topic><topic>Steel frames</topic><topic>Steel structures</topic><topic>Structural damage</topic><topic>Structural health monitoring</topic><topic>Structural system identification</topic><topic>Vibration analysis</topic><topic>Vibration control</topic><topic>Vibration monitoring</topic><topic>Vibrations</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Perez-Ramirez, Carlos A.</creatorcontrib><creatorcontrib>Amezquita-Sanchez, Juan P.</creatorcontrib><creatorcontrib>Valtierra-Rodriguez, Martin</creatorcontrib><creatorcontrib>Adeli, Hojjat</creatorcontrib><creatorcontrib>Dominguez-Gonzalez, Aurelio</creatorcontrib><creatorcontrib>Romero-Troncoso, Rene J.</creatorcontrib><collection>CrossRef</collection><collection>Engineered Materials Abstracts</collection><collection>Environment Abstracts</collection><collection>METADEX</collection><collection>Technology Research Database</collection><collection>Environmental Sciences and Pollution Management</collection><collection>Engineering Research Database</collection><collection>Materials Research Database</collection><collection>Civil Engineering Abstracts</collection><collection>Environment Abstracts</collection><jtitle>Engineering structures</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Perez-Ramirez, Carlos A.</au><au>Amezquita-Sanchez, Juan P.</au><au>Valtierra-Rodriguez, Martin</au><au>Adeli, Hojjat</au><au>Dominguez-Gonzalez, Aurelio</au><au>Romero-Troncoso, Rene J.</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Recurrent neural network model with Bayesian training and mutual information for response prediction of large buildings</atitle><jtitle>Engineering structures</jtitle><date>2019-01-01</date><risdate>2019</risdate><volume>178</volume><spage>603</spage><epage>615</epage><pages>603-615</pages><issn>0141-0296</issn><eissn>1873-7323</eissn><abstract>•A new methodology for accurate response prediction of large structures is proposed.•It uses EMD, MI index, and a probabilistic Bayesian-based training algorithm.•An MI index is proposed to determine the optimum number of neurons in the NN model.•Bayesian regularization is proposed to train the optimized NN model.•It is applied to a 1:20-scaled 38-story highrise building and a 5-story steel frame.
An accurate response prediction model is of great importance in various applications such as damage detection, structural health monitoring, and vibration control. Development of such a methodology for large civil structures is challenging because of their size and complicated behavior and noise-contaminated, nonlinear, and nonstationary nature of the signals. In addition, the prediction model must have a low computational burden for real-time applications. In this article, a new methodology and a nonlinear autoregressive exogenous model (NARX)-based recurrent neural network (NN) model is presented for accurate response prediction of large structures. The methodology is based on adroit integration of three concepts: a recent signal processing concept, empirical mode decomposition (EMD), mutual information (MI) index from the information theory, and a probabilistic Bayesian-based training algorithm. The EMD method is used to remove the noise in the measured signals. An MI index is proposed to determine the optimum number of neurons in the hidden layer of the NN model with the goal of reducing the computational requirements without affecting its performance. Finally, Bayesian regularization (BR) is proposed to train the optimized NN model. The effectiveness of the proposed methodology is assessed by predicting the structural response of a 1:20-scaled 38-story highrise building structure subjected to seismic excitations and ambient vibrations, and a five-story steel frame subjected to different levels of the Kobe earthquake.</abstract><cop>Kidlington</cop><pub>Elsevier Ltd</pub><doi>10.1016/j.engstruct.2018.10.065</doi><tpages>13</tpages><orcidid>https://orcid.org/0000-0003-3839-1396</orcidid><orcidid>https://orcid.org/0000-0001-5718-1453</orcidid></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 0141-0296 |
| ispartof | Engineering structures, 2019-01, Vol.178, p.603-615 |
| issn | 0141-0296 1873-7323 |
| language | eng |
| recordid | cdi_proquest_journals_2166740242 |
| source | ScienceDirect Journals (5 years ago - present) |
| subjects | Aseismic buildings Autoregressive models Bayesian analysis Computation Computational mathematics Damage detection Data processing Earthquake damage Earthquakes Empirical mode decomposition High rise buildings Highrise building structures Information processing Information theory Methodology Neural networks Non-linear autoregressive exogenous model Prediction models Recurrent neural networks Regularization Seismic activity Seismic engineering Signal processing Steel frames Steel structures Structural damage Structural health monitoring Structural system identification Vibration analysis Vibration control Vibration monitoring Vibrations |
| title | Recurrent neural network model with Bayesian training and mutual information for response prediction of large buildings |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-06T12%3A24%3A09IST&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=Recurrent%20neural%20network%20model%20with%20Bayesian%20training%20and%20mutual%20information%20for%20response%20prediction%20of%20large%20buildings&rft.jtitle=Engineering%20structures&rft.au=Perez-Ramirez,%20Carlos%20A.&rft.date=2019-01-01&rft.volume=178&rft.spage=603&rft.epage=615&rft.pages=603-615&rft.issn=0141-0296&rft.eissn=1873-7323&rft_id=info:doi/10.1016/j.engstruct.2018.10.065&rft_dat=%3Cproquest_cross%3E2166740242%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=2166740242&rft_id=info:pmid/&rft_els_id=S0141029618307235&rfr_iscdi=true |