黄土路基震陷系数预测的模糊神经网络方法  

Application of fuzzy neural networks for predicting seismic subsidence coefficient of loess subgrade

作　　者：谷天峰[1] 王家鼎[1] 韩晓萌[1] 郭乐[1] 

机构地区：[1]西北大学地质学系/大陆动力学国家重点试验室,陕西西安710069

出　　处：《地下水》2009年第2期97-99,119,共4页Ground water

基　　金：国家自然科学基金(40572157);铁道部重大科研项目(2005K001-A(G))

摘　　要：以郑(州)-西(安)高速铁路为例,采集黄土样品,进行室内土工试验及动三轴震陷实验,以影响震陷系数主要因素为基础,建立了高速铁路黄土路基震陷系数的模糊神经网络模型,通过对样本的训练和预测,表明该模型预测的结果与动三轴试验震陷结果比较接近,是一种比较理想的预测方法。Taking Zhengzhou -Xi＇an express railway as an example, loess as a sample, the article carries out the dynamic experiments. Based on the analysis of the main influencing factors, a fuzzy neural networks model for predicting seismic subsidence coefficient of loess subgrade has been established. After trained by the samples, the prediction results are close to those gained by the dynamic triaxial test, so it shows an ideal method for forecasting.

关 键 词：模糊神经网络 黄土路基 震陷系数 动三轴 

分 类 号：P642[天文地球—工程地质学]