资源描述:
《以径向基函数类神经网络预测承受纯扭力钢筋混凝土梁之扭力强度》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、以径向基函数类神经网络预测承受纯扭力钢筋混凝土梁之扭力强度汤兆纬1颜聪2正修科技大学土木系副教授2国立中兴大学土木工程系教授关键词:钢筋混凝土梁,扭力强度,径向基函数网络。1摘要除了挠曲力、剪力、轴向压力与拉力之外,扭力也是结构上一种基本的作用力。混凝土构件的扭力破坏是因其拉应力超过允许值所造成,而拉应力则系由扭力所引致的纯剪力受力状态。因此,扭力强度为混凝土的一种重要力学性质,在各建筑与桥梁设计规范中均须纳入考虑。然而,混凝土于扭力作用下的非线性行为相当复杂,其数理模式不易建立。有鉴于现今实验数据搜集的便利及数据分析技术的改善,研发容易、方便
2、使用且准确的混凝土扭力强度预测方法将是一件有意义的事。本文首先搜集承受纯扭力作用之矩形断面钢筋混凝土梁扭力强度数据,以免除繁复的试验工作;其次,建构径向基函数网络(radialbasisfunctionnetworks,简称RBFN),以预测含腹筋钢筋混凝土梁的极限扭力强度,并将所建构RBFN评估模式之预测值与现有钢筋混凝土梁扭力分析模式之预测值作比较。研究结果显示,应用RBFN可有效预测含腹筋钢筋混凝土梁的扭力强度,且其预测值的准确性也比既有经验公式来得精确。MODELINGTORSIONALSTRENGTHOFREINFORCEDCONCR
3、ETEBEAMSSUBJECTEDTOPURETORSIONUSINGRADIALBASISFUNCTIONNEURALNETWORKSChao-WeiTang1TsongYen2DepartmentofCivilEngineering,Cheng-ShiuUniversity,KaohsiungCounty,Taiwan,R.O.C.2DepartmentofCivilEngineering,NationalChung-HsingUniversity,Taichung,Taiwan,R.O.C.KeyWords:reinforcedconcr
4、etebeam;torsionalstrength;radialbasisfunctionnetwork.1ABSTRACTBesidesflexure,shearandaxialcompression/tension,torsionalsoformsoneofthebasicstructuralactions.Torsionalfailureofconcretemembersisinitiatedbythetensilestressdevelopedduetoastateofpureshear,whicharisesduetotorsion.
5、Therefore,torsionalstrengthisoneofthecriticalconcretemechanicalpropertiesthatareindispensablyusedindifferentbuildingandbridgedesigncodes.However,thenonlinearbehaviorofconcreteundertorsionisverycomplicated;modelingitsbehaviorisahardtask.Thus,itwouldbeofinteresttodevelopnewmet
6、hodsthatareeasier,convenient,andaccuratethantheexistingmethodsinlightoftheavailabilityofmoreexperimentaldataandrecentadvanceintheareaofdataanalysistechniques.Inthisstudy,adatabaseontorsionalfailureofRCbeamswithrectangularsectionsubjectedtopuretorsionwasretrievedfromtheexisti
7、ngliteratureforanalysisinsteadofthepracticalandexperimentaldata.Radialbasisfunctionnetworks(RBFN)aredevelopedsequentiallyandtheultimatetorsionalstrengthofeachbeamisdeterminedfromtheRBFNmodel.Besides,theRBFNmodel’spredictionsforbothtrainingandtestdatawerealsocomparedtothoseob
8、tainedusingempiricalequations.ItwasfoundthattheRBFNmodelcouldinfersolutions