{"title":"基于集成机器学习模型的结构地震响应快速估计框架","authors":"Chunxiang Li, Hai Li, Xu Chen","doi":"10.12989/SSS.2021.28.3.425","DOIUrl":null,"url":null,"abstract":"While recognized as most rigorous procedure leading to 'exact' structural seismic responses, nonlinear time history analysis is usually time consuming and computational demanding, especially when numerous structures remain to be analyzed. This paper proposes a framework to improve the time efficiency in evaluating the structural seismic demands, using ensemble machine learning models based on 'classification-regression' philosophy. Typical tall pier bridges widely located in southwest China are employed as illustrative examples to validate the efficiency and performance of this proposed framework. The results and discussion show that with properly selected input variables, the proposed ensemble model (ORF-ANN herein) performs better in predicting seismic demands than other single learning algorithms (i.e., ANN and ORF), while the time efficiency is improved over 90%. This proposed model could drastically improve the efficiency for determining structural parameters in preliminary design process, and thus reduce the iterations of trail analysis. Additionally, the model constructed from proposed framework is believed especially favored for evaluating the post-earthquake states/resilience of a region and/or highway network, where thousands of structures might be contained, and conducting nonlinear time history analysis for each one would be prohibitively time consuming and delay the rescue operations.","PeriodicalId":2,"journal":{"name":"ACS Applied Bio Materials","volume":null,"pages":null},"PeriodicalIF":4.6000,"publicationDate":"2021-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"10","resultStr":"{\"title\":\"A framework for fast estimation of structural seismic responses using ensemble machine learning model\",\"authors\":\"Chunxiang Li, Hai Li, Xu Chen\",\"doi\":\"10.12989/SSS.2021.28.3.425\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"While recognized as most rigorous procedure leading to 'exact' structural seismic responses, nonlinear time history analysis is usually time consuming and computational demanding, especially when numerous structures remain to be analyzed. This paper proposes a framework to improve the time efficiency in evaluating the structural seismic demands, using ensemble machine learning models based on 'classification-regression' philosophy. Typical tall pier bridges widely located in southwest China are employed as illustrative examples to validate the efficiency and performance of this proposed framework. The results and discussion show that with properly selected input variables, the proposed ensemble model (ORF-ANN herein) performs better in predicting seismic demands than other single learning algorithms (i.e., ANN and ORF), while the time efficiency is improved over 90%. This proposed model could drastically improve the efficiency for determining structural parameters in preliminary design process, and thus reduce the iterations of trail analysis. Additionally, the model constructed from proposed framework is believed especially favored for evaluating the post-earthquake states/resilience of a region and/or highway network, where thousands of structures might be contained, and conducting nonlinear time history analysis for each one would be prohibitively time consuming and delay the rescue operations.\",\"PeriodicalId\":2,\"journal\":{\"name\":\"ACS Applied Bio Materials\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":4.6000,\"publicationDate\":\"2021-09-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"10\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"ACS Applied Bio Materials\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://doi.org/10.12989/SSS.2021.28.3.425\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"MATERIALS SCIENCE, BIOMATERIALS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACS Applied Bio Materials","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.12989/SSS.2021.28.3.425","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MATERIALS SCIENCE, BIOMATERIALS","Score":null,"Total":0}
A framework for fast estimation of structural seismic responses using ensemble machine learning model
While recognized as most rigorous procedure leading to 'exact' structural seismic responses, nonlinear time history analysis is usually time consuming and computational demanding, especially when numerous structures remain to be analyzed. This paper proposes a framework to improve the time efficiency in evaluating the structural seismic demands, using ensemble machine learning models based on 'classification-regression' philosophy. Typical tall pier bridges widely located in southwest China are employed as illustrative examples to validate the efficiency and performance of this proposed framework. The results and discussion show that with properly selected input variables, the proposed ensemble model (ORF-ANN herein) performs better in predicting seismic demands than other single learning algorithms (i.e., ANN and ORF), while the time efficiency is improved over 90%. This proposed model could drastically improve the efficiency for determining structural parameters in preliminary design process, and thus reduce the iterations of trail analysis. Additionally, the model constructed from proposed framework is believed especially favored for evaluating the post-earthquake states/resilience of a region and/or highway network, where thousands of structures might be contained, and conducting nonlinear time history analysis for each one would be prohibitively time consuming and delay the rescue operations.