|Table of Contents|

[1] Hu Wusheng, Sun Lu,. Neural network based method for compensating model error [J]. Journal of Southeast University (English Edition), 2009, 25 (3): 400-403. [doi:10.3969/j.issn.1003-7985.2009.03.024]
Copy

Neural network based method for compensating model error()
基于神经网络方法的模型误差补偿
Share:

Journal of Southeast University (English Edition)[ISSN:1003-7985/CN:32-1325/N]

Volumn:
25
Issue:
2009 3
Page:
400-403
Research Field:
Surveying and Mapping and Navigation
Publishing date:
2009-09-30

Info

Title:
Neural network based method for compensating model error
基于神经网络方法的模型误差补偿
Author(s):
Hu Wusheng1 2 Sun Lu2
1 School of Transportation, Southeast University, Nanjing 210096, China
2 School of Civil Engineering, Catholic University of America, Washington DC 20064, USA
胡伍生1 2 孙璐2
1东南大学交通学院, 南京 210096; 2美国华盛顿天主教大学土木工程学院, 华盛顿 20064
Keywords:
model error neural network BP algorithm compensating
模型误差 神经网络 BP算法 补偿
PACS:
P207
DOI:
10.3969/j.issn.1003-7985.2009.03.024
Abstract:
Two traditional methods for compensating function model errors, the method of adding systematic parameters and the least-squares collection method, are introduced.A proposed method based on a BP neural network(called the H-BP algorithm)for compensating function model errors is put forward.The function model is assumed as y=f(x11, x22, …, xn), and the special structure of the H-BP algorithm is determined as(n+1)×p×1, where(n+1)is the element number of the input layer, and the elements are x11, x22, …, xn and y′(y′ is the value calculated by the function model); p is the element number of the hidden layer, and it is usually determined after many tests;1 is the element number of the output layer, and the element is Δy=y0-y′0(y0 is the known value of the sample).The calculation steps of the H-BP algorithm are introduced in detail.And then, the results of three methods for compensating function model errors from one engineering project are compared with each other.After being compensated, the accuracy of the traditional methods is about ±19 mm, and the accuracy of the H-BP algorithm is ±4.3 mm. It shows that the proposed method based on a neural network is more effective than traditional methods for compensating function model errors.
介绍了2种补偿模型误差的传统方法:附加系统参数方法和最小二乘配置法.提出了一种基于BP算法的补偿模型误差的神经网络方法, 简称为H-BP算法.假设函数模型为y=f(x11, x22, …, xn), 则H-BP算法的神经网络结构为(n+1)×p×1, (n+1)是输入层元素个数, 具体为x11, x22, …, xny′, 其中y′是函数模型计算值;p为隐含层节点数, 一般通过大量试验得到;1是输出层元素个数, 具体为Δy=y0-y′0, 其中y00是样本真值.然后, 详细介绍了H-BP算法的具体计算步骤.最后, 结合一个工程实例, 对3种补偿方法的结果进行了详细对比分析.传统方法补偿之后的精度约为±19 mm, H-BP算法补偿之后的精度为±4.3 mm.结果表明, 与传统方法相比, 新方法对模型误差的补偿效果更好.

References:

[1] Tao Benzao. Statistical analyses of surveying data [M].Beijing:Surveying and Mapping Press, 1992:55-87.(in Chinese)
[2] Baarda W.A testing procedure for use in geodetic networks[J]. Neth Geod Comm Publ on Geodesy, 1968, 2(5):27-55.
[3] Craven P, Wahba G.Smoothing noisy data with spline functions [J].Numer Math, 1979, 31(2):377-403.
[4] Cook R D, Weisberg S.Residuals and influence in regression [M].London:Chapmanand Hall Press, 1982:23-41.
[5] Aduol F W O.Geodetic parameter estimation through iterative weighting[J].Survey Review, 1994, 32(252):359-367.
[6] Marshall J, Bethel J.Basic concepts of L1 norm minimization for surveying applications [J].Journal of Surveying Engineering, 1996, 122(4):168-179.
[7] Ding X, Colemen R.Multiple outlier detection by evaluating redundancy contributions of observations [J].Journal of Geodesy, 1996, 70(8):489-498.
[8] Fischer B, Hegland M.Collocation filtering and nonparametric regression [J].Zeitschrift für Vermessungswesen, 1999, 124(1):17-24.
[9] Hu Wusheng, Sha Yuejin, Kuang Shanlong.New method for transforming global positioning system height into normal height based on neural network [J].Journal of Surveying Engineering, 2004, 130(1):36-39.
[10] Kuhar M, Stopar B, Turk G, et al.The use of artificial neural network in geoid surface approximation[J].Allgemeine Vermessungs-Nachrichten, 2001, 108(1):22-27.
[11] Kavzoglu T, Saka M H.Modeling local GPS/leveling geoid undulations using artificial neural networks [J].J Geodesy, 2005, 78(9):520-527.
[12] Lin Laosheng.Application of a back-propagation artificial neural network to regional grid-based geoid model generation using GPS and leveling data [J].Journal of Surveying Engineering, 2007, 133(2):81-89.
[13] Hu Wusheng.The theory of neural network and its applications in engineering [M].Beijing:Surveying and Mapping Press, 2006:63-92.(in Chinese)
[14] Rao M A, Srinivas J.Neural networks:algorithms and applications [M].Pangbourne:Alpha Science International Ltd, 2003:17-113.

Memo

Memo:
Biography: Hu Wusheng(1965—), male, doctor, professor, wusheng.hu@163.com.
Foundation items: The National Basic Research Program of China(973 Program)(No.2006CB705501), the National High Technology Research and Development Program of China(863 Program)(No.2007AA12Z228).
Citation: Hu Wusheng, Sun Lu.Neural network based method for compensating model error[J].Journal of Southeast University(English Edition), 2009, 25(3):400-403.
Last Update: 2009-09-20