基于人工神经网络的大坝变形分析与预报--以西津大坝27#点的变形监测为例

基于人工神经网络的大坝变形分析与预报--以西津大坝27^#点的变形监测为例

马丽霞¹|王凤艳¹,陈剑平²

1.吉林大学地球探测科学与技术学院|长春 130026;2.吉林大学建设工程学院|长春 130026

收稿日期:2008-10-23 出版日期:2009-05-26 发布日期:2009-05-26
通讯作者: 王凤艳（1970-），女，吉林公主岭人，副教授，主要从事工程测量及地质灾害的研究,Tel:0431-88502313 E-mail:wangfy@jlu.edu.cn
作者简介:马丽霞(1985-)|女|山西长治人|硕士研究生|主要从事大地测量和工程测量方面的研究|E-mail:immlx85@gmail.com
基金资助:
国家自然科学基金项目（40872170）

Analysis &|Prediction of Dam Deformation Based on ANN-An Example of Deformation at Monitoring Point 27 of Xijin Dam

MA Li-xia¹,WANG Feng-yan¹|CHEN Jian-ping²

1.College of GeoExploration of Science and Technology, Jilin University, Changchun 130026, China;2.College of Construction Engineering, Jilin University, Changchun 130026, China

Received:2008-10-23 Online:2009-05-26 Published:2009-05-26

摘要/Abstract

摘要：

以MATLAB语言为基础，应用BP神经网络、逐步回归分析进行西津大坝27^#点的变形分析与预报研究。在此基础上，进一步提出了逐步回归BP神经网络组合的预报方法，并探讨了3种方法的预报结果。研究表明，BP神经网络用于大坝变形分析与预报是可行的，所提出的逐步回归BP神经网络组合法提高了变形影响因子选择的科学性，在预报效果上，优于前两种方法。

关键词: 大坝变形预报, BP网络, 逐步回归

Abstract:

On the basis of MATLAB programming language, BP neural network and stepwise regression are applied to predict the deformation at monitoring point No. 27 of Xijin dam. Furthermore, the combination prediction method of BP network and stepwise regression is put forward and the prediction results of 3 methods are compared. The research shows that it is feasible to apply BP neural network to dam deformation analysis and prediction. In addition, the combination method improves the scientific quality to choice deformation factors. The combination method is prior to the method of stepwise regression in prediction result and it is worth to further study.

Key words: dam deformation prediction, BP network, stepwise regression

中图分类号:

P642.3

马丽霞, 王凤艳, 陈剑平. 基于人工神经网络的大坝变形分析与预报--以西津大坝27^#点的变形监测为例[J]. J4, 2009, 39(3): 487-491.

MA Li-xia, WANG Feng-yan, CHEN Jian-ping. Analysis &|Prediction of Dam Deformation Based on ANN-An Example of Deformation at Monitoring Point 27 of Xijin Dam[J]. J4, 2009, 39(3): 487-491.

参考文献

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed