%0 Journal Article
%A GOU Li-jie
%A LIU Jia-shun
%T Application of RBF Neural Network Model to Evaluating Sand Liquefaction
%D 2013
%R 10.3969/j.issn.1001-5485.2013.05.017
%J Journal of Yangtze River Scientific Research Institute
%P 76-81
%V 30
%N 5
%X The neural network toolbox of MATLAB7.0 was used to train and test some sample data of sand liquefaction collected by Tokimatsu Kohji. Eight eigenvectors clay content (<i>ρ</i><sub>c</sub>),relative compaction(<i>D</i><sub>r</sub>),critical depth of soil layer(<i>d</i><sub>s</sub>),vertical effective stress(<i>σ</i>′),groundwater level(<i>d</i><sub>w</sub>),magnitude of earthquake(<i>M</i>),maximum horizontal ground acceleration(<i>α</i><sub>max</sub>) and standard penetration number(SPT-<i>N</i>) were selected as input parameters of the RBF neural network. Furthermore,the established RBF neural network model was used to analyze the effect of each factor on the sand liquefaction. Results of the relative contribution of each factor showed that <i>α</i><sub>max</sub> was the biggest influencing factor on the evaluation index of sand liquefaction,followed by SPT-<i>N</i> and <i>d</i><sub>w</sub>. The evaluation index increased with the rise of <i>α</i><sub>max</sub>,while reduced with the increase of SPT-<i>N</i> and <i>d</i><sub>w</sub>. The evaluation index shows a logarithmic relation with <i>α</i><sub>max</sub>,cubic polynomial relation with SPT-<i>N</i>,and a negative linear relation with <i>d</i><sub>w</sub> . It’s revealed that the established RBF network model fully meets the requirement of evaluation accuracy for sand liquefaction. It can simulate the complex nonlinear mapping relation between the input and output data and also gives high prediction precision.
%U http://ckyyb.crsri.cn/EN/10.3969/j.issn.1001-5485.2013.05.017