摘要
针对混凝土大坝坝体和岩基参数的区间不确定性,构造具有区间分析功能的RNN(粗糙神经网络)模型,并运用该模型反演坝体和岩基区间参数值。应用区间有限元对结构进行正分析,根据区间参数反演的需要选取相应的区间学习样本,利用RNN模型对样本进行模式学习直至网络收敛,最后通过网络回想和反归一法得到坝体和岩基力学参数的区间值。研究结果表明,该方法可用于反演混凝土坝坝体和岩基区间力学参数,反演得到的区间参数值是合理的。此外,基于RNN模型的区间参数反演方法经过一定的拓展和改进,理论上可应用于反演其他类型的区间参数。
In light of the uncertainty of concrete dam and rock foundation parameters,a rough neural network （RNN） model is constructed and applied to interval parameters inversion of dam body and rock foundation.The RNN model is a combination of rough sets and BP-NN,which has the interval input and output just as rough sets and has a similar network structure to BP-NN.The steps of this method are as follows.Firstly,the structure is analyzed with interval finite element method.Secondly,the corresponding interval samples are chosen from interval FEM results according to special requirements for interval parameters inversion.Thirdly,the interval samples are trained continuously by RNN model until the error is less than a given threshold.Lastly,the interval parameters of dam body and rock foundation are calculated by means of network recollection and back normalization.It is shown that this method can be used to interval mechanical parameters inversion of dam body and rock foundation and the result is reasonable.Moreover,this interval analysis method based on RNN model theoretically can also be used to other interval parameters inversion after some expansion and improvement.
出处
《岩土力学》
EI
CAS
CSCD
北大核心
2011年第2期547-552,共6页
Rock and Soil Mechanics
基金
湖南省教育厅项目(No.09C087)
国家自然科学基金(No.50809025)
国家科技支撑计划课题(No.2008BAB29B03)
国家重点实验室专项经费资助项目(No.2009586912)
长沙理工大学人才引进基金资助项目(No.1004271)
关键词
粗糙神经网络
区间有限元
区间反分析
不确定性参数
rough neural network
interval finite elements
interval back analysis
uncertain parameters
