Conversion of GPS Height Based on Bayesian Regularization BP Neural Network
-
摘要: 为了改善BP神经网络在GPS高程转换过程中过拟合的现象,提出了用贝叶斯正则化算法的BP神经网络转换GPS高程的新方法,并利用区域GPS/水准数据,将新方法和未采用正则化算法的BP神经网络进行GPS高程转换的比较.结果表明:在较大区域和高程异常呈不规则的情况下,新方法不仅可以有效提高GPS高程转换的精度,而且通过贝叶斯正则化算法可以改善网络结构,抑制过拟合现象.在约10 km的GPS基线尺度上,新方法可以得到精度达0.050 m的正常高.Abstract: In order to improve the over-fitting in GPS(global positioning system) height conversion using BP(back propagation) neural network,a new method of GPS height conversion based on the Bayesian regularization BP neural network was proposed.Using the GPS/leveling data in a certain area,this new method was compared with the BP neural network without using the regularization algorithm for GPS height conversion.The research results show that the new method can not only improve the precision of GPS height conversion but also restrain the over-fitting through using the Bayesian regularization algorithm to improve the structure of neural networks in cases with a big area and anomalous height anomaly.The precision of GPS height conversion can achieve 0.050 m to an about 10 km baseline with the new method.
-
刘成龙,杨天宇.基于BP神经网络的GPS高程拟合方法的探讨[J].西南交通大学学报,2007,42(2):148-152.LIU Chenglong,YANG Tianyu.Study on method of GPS height fitting based on BP artificial neural network[J].Journal of Southwest Jiaotong University,2007,42 (2):148-152.[2] 鲁铁定,钟小威.基于改进BP学习算法的GPS高程转换[J].测绘通报,2005(12):13-15.LU Tieding,ZHONG Xiaowei.GPS height conversion on the improvement of BP learning algorithm[J].Bulletin of Surveying and Mapping,2005(12):13-15.[3] 杨明清,靳蕃,朱达成,等.用神经网络方法转换GPS高程[J].测绘学报,1999,28(4):301-307.YANG Mingqing,JIN Fan,ZHU Daeheng,et al.Conversion of GPS height by artificial neural network method[J].Aeta Geodaetica et Cartographiea Sinica,1999,28 (4):301-307.[4] KAVZOGLU T,SAKA M H.Modeling local GPS/leveling geoid undulations using artificial neural networks[J].Journal of Geodesy,2004,78(9).520-527.[5] 胡伍生,华锡生.平坦地区转换GPS高程的混合转换方法[J].测绘学报,2002,31(2):128-133.HU Wusheng,HUA Xisheng.Conversion of GPS height in plainness area by the CF&NNM method[J].Acta Geodaetica et Cartographica Sinica,2002,31 (2):128-133.[6] 魏海坤,徐嗣鑫,宋文忠.神经网络的泛化理论和泛化方法[J].自动化学报,2001,27(6):806-813.WEI Haikun,XU Sixin,SONG Wenzhong.Generalization theory and generalization methods for neural networks[J].Aeta Automatica Siniea,2001,27(6):806-813.[7] 鲁铁定,周世健,臧德彦.关于BP神经网络转换GPS高程的若干问题[J].测绘通报,2003(8):7-9.LU Tieding,ZHOU Shijian,ZANG Deyan.The problems of GPS height conversion by artificial neural network method[J].Bulletin of Surveying and Mapping,2003(8):7-9.[8] HORMIK K.Approximation capability of multiplayer feed forward network[J].Neural Networks,1991 (4):251-257.[9] MARTIN T H,HOWARD B D,MARK H B.神经网络设计[M].戴葵,宋辉,潭明锋,等译.北京:机械工业出版社,2005:197-257.[10] 李红霞,许士国,范垂仁.基于贝叶斯正则化神经网络的径流长期预报[J].大连理工大学学报,2006,46(增刊):174-177.LI Hongxia,XU Shiguo,FAN Chuiren.Long-term prediction of runoff based on Bayesian regulation neural network[J].Journal of Dalian University of Technology,2006,46 (Sup.):174-177.[11] MACKAY D J C.Bayesian interpolation[J].Neural Computation,1992,4(3):415-447.[12] 晁定波.关于我国似大地水准面的精化及有关问题[J].武汉大学学报:信息科学版,2003,28(增刊):110-114.CHAO Dingbo.Refinement of quasi-gecid in China and relevant problems[J].Geomaties and Information Science of Wuhan University,2003,28 (Sup.):110-114.
点击查看大图
计量
- 文章访问数: 1610
- HTML全文浏览量: 62
- PDF下载量: 392
- 被引次数: 0