粘性土P-Y曲线中Y_(50)的神经网络预测  

Neural Network Forecasted Y_(50) on P-Y Curves in Clayey Soil

作　　者：谢耀峰[1] 王云球[2] 

机构地区：[1]东南大学港口航道工程系,江苏南京210096 [2]河海大学交通与海洋工程学院,江苏南京210098

出　　处：《水运工程》2005年第1期7-12,共6页Port & Waterway Engineering

摘　　要：Y50是桩粘性土P-Y曲线的一个重要参数,Y50受诸多因素的影响,随机性较大。对港口工程中常见的粘性土8个常规物理性质指标进行主成分分析,研究用若干指标判定粘性土土性的合理性。抓住三个主要影响因素桩径、不排水抗剪强度Cu与土的压缩系数αv,进行BP网络的结构设计,用神经网络原理对粘土P-Y曲线中Y50进行预测。预测结果能够满足工程精度要求。Y50 is an important parameter for piles P-Y curves in clayey soil,which has great randomicity due to many factors. The principal component analysis method studied 8 physical parameters and artificial neural network forecasted Y50 in clayey soil are researched. The BP network used for forecasting Y50 of P-Y curves in clay is designed by three main influential factors, i.e. pile-diameter,undrained shear strength Cu and compressibility coefficient of the soils ?琢v. These results, which have practical value in piles calculation, are enough to meet piles calculation requirement.

关 键 词：神经网络 Y50 主成分分析法 

分 类 号：U655.54[交通运输工程—港口、海岸及近海工程]