摘要
针对地质勘查中,土的力学参数的确定及土的分类这两类复杂问题,根据反问题理论的基本原理,提出了一种基于回归分析与RBF神经网络结合的新型智能方法,建立了从土的力学参数估计到模型分类的完整智能化分析系统。考虑到土的物理参数测定方法比较简单,且实测变异性小,而力学参数实测变异性大的特点,利用RBF神经网络的数值逼近的特性,建立了神经网络模型来逼近两者之间的函数关系,可以有效地反演力学参数。同时,利用RBF神经网络所具有的模式识别功能,为地质勘察中土层划分提供依据。通过对黄石地区岩土勘查资料的分析与预测表明,该方法简捷有效。
Mechanical parameters estimation and classification of soils are very important in geologic examination. On the basis of inverse problem theory, a new intelligent method combining RBF neural network and regression analysis is proposed. Then an intellectualized simulation system of soil is established, consisting of two neural networks for mechanical parameter estimation and model recognition. In the system, considering variability of physical parameters is much smaller than mechanical parameters of soils, an artificial neural network model is established to approach the function relationship of the two kinds of parameters. It is effective to reflect mechanical parameters according physical parameters. The mechanical parameters will be input vectors applied to the other network. Then, the new neural network is established; it can offer a good approach to soil classification. This intellectualized simulation system is applied to analyzing geologic examination data in Huangshi; and the results show that the method is simple and effective.
出处
《岩土力学》
EI
CAS
CSCD
北大核心
2007年第4期807-811,共5页
Rock and Soil Mechanics
关键词
反问题
回归分析
RBF神经网络
力学参数估计
土层分类
inverse problem
regression analysis
RBF neural network
mechanical parameters estimation
soil classification
