摘要
研究目的:影响CFG桩复合地基承载力的主要因素有桩的参数、置换率、土的物理力学特性、褥垫层厚度和施工工艺等,且各因素之间存在高度复杂的非线性关系,CFG桩复合地基的承载力比较难于确定。为合理准确预测CFG桩复合地基承载力,通过研究提出基于自适应模糊神经网络的预测方法。研究结论:在分析自适应模糊神经网络原理及结构的基础上,利用减法聚类获得模糊推理规则数目,确定网络结构,建立适用于CFG桩复合地基承载力预测的自适应模糊神经网络模型。通过对实测资料的预测结果表明,自适应模糊神经网络比BP网络和最小二乘支持向量机LS_SVM模型具有更高的精度和适应性,为CFG桩复合地基承载力的判别提供了一条新的途径。
Research purposes： The main factors influencing the bearing capacity of CFG pile composite foundation are the parameters of CFG pile; replacement rate, physical mechanical characteristics of soil, cushion thickness and construction technique, and the various factors mentioned above have highly complex non - linear relationship, so it is rather difficult to determine the bearing capacity of CFG pile composite foundation. To reasonably and accurately predict the CFG pile composite foundation ＂s bearing capacity, the adaptive - network based on fuzzy inference system was proposed. Research conclusions： The theory of adaptive - network - based fuzzy inference systems （ANFIS） was introduced and subtractive cluster method was applied to deciding fuzzy inference rules. After that, an ANFIS was used to predict the bearing capacity of CFG pile composite foundation. The predicted results show that the adaptive fuzzy neural network has higher accuracy and adaptability than the BP network and least squares support vector machines LS - SVM model, and it provides a new approach to predict the bearing capacity of CFG pile composite foundation.
出处
《铁道工程学报》
EI
北大核心
2010年第6期42-47,共6页
Journal of Railway Engineering Society
关键词
CFG桩复合地基
承载力
自适应模糊神经网络
最小二乘支持向量机
CFG pile composite foundation
bearing capacity
adaptive - network - based fuzzy inference systems
least squares support vector machine
