摘要
滑带土原位直剪试点的物质构成、含水率、应力历史均具有随机性,预剪面边界约束、恢复初始固结状态的程度、剪切过程中的排水条件等因素则具有模糊性,这些不确定性因素导致试验成果离散、代表性受到局限。以原位直剪强度参数以及破坏面现场定性地质描述、含水率测试、颗粒分析结果训练 BP 网络,利用 BP 网络的高度容错性和强大的非线性映射能力,估计滑带土的抗剪强度参数,可在一定程度上可克服试验成果离散的影响。
For in-situ shear test on soil in slide zone,substance constituent,water content and stress history of its samples are all random to some extent;boundary restrain,restoring degree to original consolidation state and draining condition in the course of shearing are fuzzy.All these uncertain factors cause test results become variant and then have no enough geological representation. Applying in-situ shear test parameters and the results of qualitative geology description of failure surface,water ratio measurement and grading analysis to train BP network;and then depending on BP network's ability of high fault-tolerance and strong nonlinear mapping,the shear strength parameters of soil in slide zone are estimated.This method could counteract the influence of the variance of test results to some extent.
出处
《岩土力学》
EI
CAS
CSCD
北大核心
2006年第S2期1145-1148,共4页
Rock and Soil Mechanics
关键词
滑带
试验条件
不确定性
抗剪强度参数
BP网络
slide zone
test condition
uncertainty
shear strength parameters
BP network