小波神经网络在高程拟合中的应用
|
摘要
基于小波神经网络对样本的学习可以准确地估计任何非线性的时间序列的优点,将其应用于GPS高程拟合,并与二次曲面模型进行了比较,试验结果表明,小波神经网络模型能够利用有限的样本信息获得最好的学习效果和泛化能力,其拟合精度优于常规二次曲面拟合方法。
The advantages of any nonlinear time series can be estimated accurately by learning samples based on wavelet neural network. It is applied to GPS height fitting and compared with the conicoid model. The experimental results show that: The wavelet neural network model can make use of the limited sample information to obtain the best learning effect and generalization ability, and its fitting precision is better than the conventional conicoid fitting method.
The advantages of any nonlinear time series can be estimated accurately by learning samples based on wavelet neural network. It is applied to GPS height fitting and compared with the conicoid model. The experimental results show that: The wavelet neural network model can make use of the limited sample information to obtain the best learning effect and generalization ability, and its fitting precision is better than the conventional conicoid fitting method.
引文
[1]刘帅,王礼江,朱建军,等.GPS高程拟合模型的优选[J].测绘工程,2006,15(4):17-19.
[2]冯光财,陈正阳.一种GPS高程拟合计算的新方法[J].矿山测量,2005(4):6-8.
[3]余宣兴,詹昊,朱明新,等.EGM2008地球重力场模型在GPS高程转换中的应用研究[J].测绘通报,2013(12):18-20.
[4]刘斌,郭际明,史俊波,等.利用EGM2008模型与地形改正进行GPS高程拟合[J].武汉大学学报(信息科学版),2016,41(4):554-558.
[5]牛志宏,宋萌勃.基于神经网络的GPS高程拟合算法探析[J].测绘技术装备,2014(2):37-40.
[6]张昊,王琪洁,朱建军,等.样本数据预处理对基于BP神经网络的GPS高程拟合的影响[J].大地测量与地球动力学,2011,31(2):125-128.
[7]付建东,吴良才.小波神经网络的改进及在GP S高程拟合中的应用研究[J].工程勘察,2015,43(2):90-93.
[8]叶发学,袁图杰,叶虎春,等.利用小波神经网络的GPS高程转换[J].测绘地理信息,2013,38(6):23-26.
[2]冯光财,陈正阳.一种GPS高程拟合计算的新方法[J].矿山测量,2005(4):6-8.
[3]余宣兴,詹昊,朱明新,等.EGM2008地球重力场模型在GPS高程转换中的应用研究[J].测绘通报,2013(12):18-20.
[4]刘斌,郭际明,史俊波,等.利用EGM2008模型与地形改正进行GPS高程拟合[J].武汉大学学报(信息科学版),2016,41(4):554-558.
[5]牛志宏,宋萌勃.基于神经网络的GPS高程拟合算法探析[J].测绘技术装备,2014(2):37-40.
[6]张昊,王琪洁,朱建军,等.样本数据预处理对基于BP神经网络的GPS高程拟合的影响[J].大地测量与地球动力学,2011,31(2):125-128.
[7]付建东,吴良才.小波神经网络的改进及在GP S高程拟合中的应用研究[J].工程勘察,2015,43(2):90-93.
[8]叶发学,袁图杰,叶虎春,等.利用小波神经网络的GPS高程转换[J].测绘地理信息,2013,38(6):23-26.