摘要
利用人工神经网络处理非线性体系的优势性,对盐渍土膨胀规律多影响因素试验数据进行了建模方法分析,提出了盐渍土盐胀率随含水量、氯化钠含量、硫酸钠含量、初始干容重和上覆荷载5因素变化的计算公式,计算结论比常规二次回归法更加符合目前对盐渍土盐胀规律的定性认识.
A non-linear neural network model is established to study the salt expansion properties of saline soil under the function of the five factors, i. e. , water content, NaC1 concentration, Na2SO4 concentration, initial dry density and overburden pressure of saline soil, based on the documents mentioned in this paper. As compared with the traditional method of quadratic stepwise regression, it shows much more advantages and creditability in solving the problem of the non-linear interaction of multi influencing factors. At the same time, the formula of counting the expansion rate of saline soil under the function of the five factors is updated and coincided the present understanding of the properties of saline soil.
出处
《冰川冻土》
CSCD
北大核心
2006年第4期607-612,共6页
Journal of Glaciology and Geocryology
关键词
人工神经网络
盐渍土
盐胀率
交互作用
artificial neural network saline soil salt expansion rate interaction
