基于加权BP神经网络的场地连续化分类
|
摘要
本文以四类标准场地模糊分类二维隶属函数为基础,以BP神经网络为手段,建立了一种新的标准场地模糊化方法,训练了现行抗震设计规范中4类标准场地的模糊化神经网络,克服了二维隶属函数在转折边界处不平缓和有突变的不足;并以所建立的4个标准场地模糊化神经网络为基础,建立了基于加权综合的场地连续化分类方法,给出了该方法分类结果的空间分布图,实现了场地分类的连续化,使场地分类结果更符合工程实际;工程应用实例表明,基于BP神经网络的场地连续化分类方法,能更合理、更客观地对工程场地进行分类,为更科学地确定地震作用和进行恰当的结构抗震设计提供了可能。
Based on the fuzzy classification method of the two-dimensional membership function and the Back Propagation(BP)network,a new fuzzy classification method for the four standard site classifications is presented.The training of fuzzy networks of four types of standard sites is fulfilled.The BP-based method overcomes the two-dimensional membership function's defect of being not gradual or gentle but involving irregular or abrupt breaks at the surfaces(junctures of planes).Based on the fuzzy classification networks of four standard sites,a weighted-integrated-based site classification method is proposed.Space distribution graphics of this method is provided.The continuous classification method is fulfilled.Finally,an engineering case is given.The results show that this method makes the classification result better meet requirements of practical construction more objectively and reasonably.Therefore,we can ascertain both the seismic action and an appropriate structural aseismic design scientifically.
Based on the fuzzy classification method of the two-dimensional membership function and the Back Propagation(BP)network,a new fuzzy classification method for the four standard site classifications is presented.The training of fuzzy networks of four types of standard sites is fulfilled.The BP-based method overcomes the two-dimensional membership function's defect of being not gradual or gentle but involving irregular or abrupt breaks at the surfaces(junctures of planes).Based on the fuzzy classification networks of four standard sites,a weighted-integrated-based site classification method is proposed.Space distribution graphics of this method is provided.The continuous classification method is fulfilled.Finally,an engineering case is given.The results show that this method makes the classification result better meet requirements of practical construction more objectively and reasonably.Therefore,we can ascertain both the seismic action and an appropriate structural aseismic design scientifically.
引文
[1]Wood HO.Dist ribution of Apparent Intensity in San Francisco1908[A].Lawson A C.The California Earth-quake of April18,1906,Report of state Earthquake Inv-setigation commission[C]Washington D C:Carnegie In-stitution of Washington,1908.220～245
[2]胡聿贤,孙平善,章在墉,田启文.场地条件对震害和地震动的影响.地震工程与工程振动,1980,1(1):34～40
[3]周锡元.土质条件对建筑物所受地震作用的影响[A].中国科学院工程力学研究所,地震工程研究报告集(第二集)[R].科学出版社,1965
[4]Seed HB,Ugas C,Lysmer J.Site Dependent Spectra forEarthquake Rseistant Dseign,Bull,Seis.Soc.Am.1976,66:221～244
[5]周锡元,王广军,苏经宇.场地分类和平均反应谱.岩土工程学报,1984,6(5):59～68
[6]王广军,苏经宇.场地类别的模糊综合评判.地震工程与工程振动,1985,5(2):28～42
[7]GB50011—2001,建筑抗震设计规范.北京:中国建筑工业出版社,2001
[8]张世海,刘叔军,欧进萍,王光远.基于二维隶属函数的场地模糊分类及其应用研究.岩土工程学报,2005,27(8):912～917
[9]GB/T50269-97,地基动力特性测试规范.北京:中国计划出版社,1998
[10]GB50191—93,构筑物抗震设计规范.北京:中国计划出版社,1994
[11]Mills P M,Zomaya A Y,Tade M O.Adaptive model-based control using neural networks control.1994,60:32～35
[12]Hecht-Nielsen R.Theory of the Back-Propagation Neural Network.Proc.IEEE International Conference on Neural Network.Washington D.C.,1989:593～605
[2]胡聿贤,孙平善,章在墉,田启文.场地条件对震害和地震动的影响.地震工程与工程振动,1980,1(1):34～40
[3]周锡元.土质条件对建筑物所受地震作用的影响[A].中国科学院工程力学研究所,地震工程研究报告集(第二集)[R].科学出版社,1965
[4]Seed HB,Ugas C,Lysmer J.Site Dependent Spectra forEarthquake Rseistant Dseign,Bull,Seis.Soc.Am.1976,66:221～244
[5]周锡元,王广军,苏经宇.场地分类和平均反应谱.岩土工程学报,1984,6(5):59～68
[6]王广军,苏经宇.场地类别的模糊综合评判.地震工程与工程振动,1985,5(2):28～42
[7]GB50011—2001,建筑抗震设计规范.北京:中国建筑工业出版社,2001
[8]张世海,刘叔军,欧进萍,王光远.基于二维隶属函数的场地模糊分类及其应用研究.岩土工程学报,2005,27(8):912～917
[9]GB/T50269-97,地基动力特性测试规范.北京:中国计划出版社,1998
[10]GB50191—93,构筑物抗震设计规范.北京:中国计划出版社,1994
[11]Mills P M,Zomaya A Y,Tade M O.Adaptive model-based control using neural networks control.1994,60:32～35
[12]Hecht-Nielsen R.Theory of the Back-Propagation Neural Network.Proc.IEEE International Conference on Neural Network.Washington D.C.,1989:593～605
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心