【摘要】 通过有限元分析对引水渠道进行结构计算时,为了使数值计算的结果更可靠,需要对相关计算参数进行反演。根据工程实际情况,首先选取多种土体结构参数的组合作为参数训练样本,采用有限元法利用不同的土体结构参数组合对渠道的沉降变形进行数值计算;基于水位变化与引水渠道土体沉降的关系,将选取的样本投入RBF(Radical Basis Function,径向基函数)神经网络中训练,建立渠道土体结构参数与因水位变化引起的渠道土体沉降值之间的映射关系;最后根据渠道实际变形监测值,采用RBF神经网络反演得到相关变形参数,以实现对引水渠道结构的精确计算。更多还原

【Abstract】 In order to make the numerical calculation results more reliable,it is necessary to invert the relevant calculation parameters when calculating the structure of the water diversion channel through finite element analysis. In the study,a variety of structural parameter samples were firstly drawn up and the settlement deformation of the channel was calculated by finite element method.Then the RBF( Radical Basis Function) neural network training sample was used to establish the mapping relationship between channel deformation parameters and channel subsidence deformation.Finally,according to the actual deformation monitoring value of the channel,RBF neural network is used to retrieve the deformation parameters,so as to achieve the accurate calculation of the structure of the diversion channel.更多还原