基于主成分分析的优化神经网络模型及静压管桩单桩极限承载力预测

›› 2011, Vol. 32 ›› Issue (S2): 634-640.

基于主成分分析的优化神经网络模型及静压管桩单桩极限承载力预测

史永强^{1, 2}，赵俭斌²，杨军²

1. 哈尔滨工业大学土木工程学院，哈尔滨 150001；2. 沈阳建筑大学土木工程学院，沈阳 110168

收稿日期:2010-03-31 出版日期:2011-08-10 发布日期:2011-08-26
作者简介:史永强，男，1982年生，博士研究生，讲师，主要从事与岩土工程相关的研究工作
基金资助:
辽宁省重点实验室项目资助（No.2008S196）；辽宁省高校重点实验室项目资助（No.YT-200903）；沈阳市科技局项目资助（No.1081271- 9-00-3）

Optimized neural network model for predicting ultimate bearing capacity of statically-pressured pipe pile based on principal component analysis

SHI Yong-qiang^{1, 2}, ZHAO Jian-bin², YANG Jun²

1. School of Civil Engineering, Harbin Institute of Technology, Harbin 150001, China; 2. School of Civil Engineering, Shenyang Jianzhu University, Shenyang 110168, China

Received:2010-03-31 Online:2011-08-10 Published:2011-08-26

摘要/Abstract

摘要： 引入主成分分析法和基于共轭梯度优化算法的人工神经网络模型原理，建立了静压管桩单桩竖向承载力预测估算的新方法。通过对影响单桩极限承载力的各因素进行主成分分析确定了综合变量，构建了以综合变量为输入，以单桩极限承载力为输出的神经网络模型。应用神经网络结构分析的共轭梯度算法，优化计算获得给定样本的网络权值和阈值，获得静压管桩极限承载力的估算网络，应用实例分析计算了静压管桩单桩极限承载力问题。结果表明，利用所建立的神经网络预测静压管桩极限承载力是可行的，且具有较好的预测精度和良好的适用性。该方法为静压管桩竖向承载性状的理论分析开辟了一个新的研究途径，为今后相关问题研究提供借鉴和指导

关键词: 静压管桩, 单桩极限承载力, 主成分分析, 共轭梯度算法, 人工神经网络模型

Abstract: A new method for predicting the vertical ultimate bearing capacity of statically-pressured pipe pile is established by using artificial neural network based on principal component analysis and conjugate gradient algorithm. The integrated variables are determined through the principal component analysis of factors which influenced the ultimate bearing capacity of single pile. Then, the artificial neural network with integrated variables as input and ultimate bearing capacity as output is confirmed. Also, the conjugate gradient algorithm is applied to computing the weight values and threshold values with given samples in order to obtain the network of predicting ultimate bearing capacity. At last, the engineering examples are used to verify the established model. It could be concluded that it is feasible to predicting the ultimate bearing capacity of statically-pressured pipe pile with this model, which has a better prediction accuracy and applicability. This method provides a new way to analyze the vertical bearing character of statically-pressured pipe pile, and gives references to the relevant problems.

Key words: statically-pressured pipe pile, ultimate bearing capacity of single pile, principal component analysis, conjugate gradient algorithm, artificial neural network

中图分类号:

TU 443

史永强，赵俭斌，杨军. 基于主成分分析的优化神经网络模型及静压管桩单桩极限承载力预测[J]. , 2011, 32(S2): 634-640.

SHI Yong-qiang , ZHAO Jian-bin , YANG Jun. Optimized neural network model for predicting ultimate bearing capacity of statically-pressured pipe pile based on principal component analysis[J]. , 2011, 32(S2): 634-640.

参考文献

相关文章 15

[1]	成帅，李术才，李利平，石少帅，周宗青，袁永才. 基于多元监测信息融合分析的突水灾害状态判识方法[J]. , 2018, 39(7): 2509-2517.
[2]	王家全，刘垒雷，朱庆盛，张昊,. 红黏土地层静压闭口管桩残余应力模型试验分析[J]. , 2017, 38(7): 1878-1886.
[3]	黄丹，李小青. 基于微裂纹发育特性的大理岩特征强度数值模拟研究[J]. , 2017, 38(1): 253-262.
[4]	钱兆明，任高峰，褚夫蛟，秦绍兵, . 基于PCA法和Fisher判别分析法的岩体质量等级分类[J]. , 2016, 37(S2): 427-432.
[5]	宫凤强，李嘉维,. 基于PCA-DDA原理的砂土液化预测模型及应用[J]. , 2016, 37(S1): 448-454.
[6]	许昌，岳东杰，董育烦，邓成发. 基于主成分和半参数的大坝变形监测回归模型[J]. , 2011, 32(12): 3738-3742.
[7]	骆行文，姚海林. 基于主成分分析的岩石质量综合评价模型与应用[J]. , 2010, 31(S2): 452-0455.
[8]	李顺群，高凌霞，柴寿喜，王英红，刘双菊. 一维固结条件下黏土微结构再造过程研究和描述[J]. , 2010, 31(11): 3435-3440.
[9]	谷琼，蔡之华，朱莉，黄波. 基于PCA-GEP算法的边坡稳定性预测[J]. , 2009, 30(3): 757-761.
[10]	许小健，钱德玲，郭文爱，王鉴. 基于RAGA的指数曲线模型预测基桩承载力[J]. , 2009, 30(1): 139-142.
[11]	王俊林，王志宽，马艳，. 静压管桩单桩极限承载力与终压力关系的探讨[J]. , 2008, 29(S1): 631-634.
[12]	关喜才. 新型固化剂ESC在搅拌桩中的应用研究[J]. , 2008, 29(11): 3149-3152.
[13]	汤皓，陈国兴，李方明. 基于ANN和GIS耦合模型的场地地震液化势评价系统研究[J]. , 2006, 27(S1): 1007-1012.
[14]	张宏，柳艳华，石名磊，赵俊明,. 海陆交互相下粘性土的性状研究[J]. , 2005, 26(9): 1491-1494.
[15]	罗战友，董清华，龚晓南. 未达到破坏的单桩极限承载力的灰色预测[J]. , 2004, 25(2): 304-315.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed