基于RBF神经网络模型的砂土液化震陷预估法
|
摘要
以我国海城地震、唐山地震和日本新澙地震中建筑物地基的液化震陷实测资料为基础,地震动方面选取地震烈度I,上部结构特征方面选取基底压力p、基础类型T、宽深比BD和建筑物的长高比L/H,地基土方面选取土的相对密度Dr、上覆非液化土层厚度Dn、地下水位dw,共8个影响建筑物地基震陷的主要因素作为神经网络模型的输入参数,地基震陷量与液化土层的深度之比sD作为神经网络模型的输出,采用径向基函数神经网络模型建立建筑物地基的液化震陷预估模型,并利用该模型建立了因素影响趋势线,通过对该神经网络模型的建立、运行和检验,得到了各因素对砂土液化引起的地基震陷量大小的若干影响规律。
Based on erthquake-induced liquefaction settlement data from Niigata earthquake in Japan,Haicheng earthquake and Tangshan earthquake in China,selecting 8 main influential indexes of the earthquake-induced liquefaction settlement of foundation,i.e.earthquake intensity for describing earthquake ground motion characteristics,average pressure of foundation base,foundation type,ratio of liquefaction soil depth to width of foundation,ratio between length and height of building for describing building characteristics,relative density of sand,non-liquefaction soil thickness covered above and underground water level for describing ground characteristics,as the input variables of neural network model,and choosing the ratio of earthquake-induced settlement to liquefaction soil depth as the output variable of neural network model,with radial basis function(RBF) neural network model,the model of earthquake-induced foundation settlement evaluation is constructed,and from which the factor-effected potential lines of earthquake-induced foundation settlement are also given.Through the construction,operation and verification of this neural network model,some influential rules of earthquake-induced liquefaction settlement of foundation are obtained.
Based on erthquake-induced liquefaction settlement data from Niigata earthquake in Japan,Haicheng earthquake and Tangshan earthquake in China,selecting 8 main influential indexes of the earthquake-induced liquefaction settlement of foundation,i.e.earthquake intensity for describing earthquake ground motion characteristics,average pressure of foundation base,foundation type,ratio of liquefaction soil depth to width of foundation,ratio between length and height of building for describing building characteristics,relative density of sand,non-liquefaction soil thickness covered above and underground water level for describing ground characteristics,as the input variables of neural network model,and choosing the ratio of earthquake-induced settlement to liquefaction soil depth as the output variable of neural network model,with radial basis function(RBF) neural network model,the model of earthquake-induced foundation settlement evaluation is constructed,and from which the factor-effected potential lines of earthquake-induced foundation settlement are also given.Through the construction,operation and verification of this neural network model,some influential rules of earthquake-induced liquefaction settlement of foundation are obtained.
引文
[1]陈国兴.高层建筑基础设计[M].北京:中国建筑工业出版社,2000.
[2]陈国兴,谢君斐,张克绪.天然地基浅基础的震害分析[J].岩土工程学报,1995,(1):66-72.
[3]郭晶,杨章玉.MATLAB6.5辅助神经网络分析与设计[M].北京:电子工业出版社,2003.
[4]刘惠珊,张在明.地震区的场地与地基基础[M].北京:中国建筑工业出版社,1994.
[5]刘颖,谢君斐.砂土震动液化[M].北京:地震出版社,1984.
[6]黄熙龄,崔竞立,顾晓鲁.天津市区地基震害概况,唐山大地震震害(一)[M].北京:地震出版社,1985.
[2]陈国兴,谢君斐,张克绪.天然地基浅基础的震害分析[J].岩土工程学报,1995,(1):66-72.
[3]郭晶,杨章玉.MATLAB6.5辅助神经网络分析与设计[M].北京:电子工业出版社,2003.
[4]刘惠珊,张在明.地震区的场地与地基基础[M].北京:中国建筑工业出版社,1994.
[5]刘颖,谢君斐.砂土震动液化[M].北京:地震出版社,1984.
[6]黄熙龄,崔竞立,顾晓鲁.天津市区地基震害概况,唐山大地震震害(一)[M].北京:地震出版社,1985.
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