摘要
Based on principal component analysis, the rules of clayey soil's behaviors affected by varied indices were studied. It was discovered that the common method of the single liquidity index IL used to determine the consistency of silt-clay or silt-loam was not rational. It was more rational that the liquidity index IL combined with the void ratio e characterized the behavior of silt-clay. Similarly the index of e depicted the nature of sandy loam more rationally than IL. The method of predicting the pile shafted resistance by the two indices of e and IL, which was more accurate, was obtained by the methodology of back propagation (BP) artificial neural networks combined with principal component analysis. It was also observed that the pile shaft resistance increased with the increase of depth within a critical affect-depth ranging from 20 to 30 m, and the harder the clayey soil consistency was, the shallower the critical depth was.
采用主成分分析方法,就粘性土多指标反映其性质的规律进行了研究.研究表明,采用液性指数作为单一指标的传统粘性土物理状态划分方法,在反映亚粘土和亚砂土性质时不尽合理.而采用液性指数IL.结合孔隙比e反映粉质粘土的特性更加合理.同样,孔隙比e比液性指数IL能更好地描述亚粘土的天然特性.采用人工神经网络结合主成分分析,得出应用孔隙比e和液性指数IL两个指标来预测桩侧摩阻力更为精确.同时发现在一定临界影响深度范围内(20-30 m),桩侧摩阻力随深度的增加而增加,且粘性土的稠度愈硬,临界深度愈浅.
