转换GPS高程的BP神经网络方法研究

对于目前大范围点位分布不均,拟合法高程转换存在效果失真、模型误差等问题,本文给出了改进的BP神经网络方法转换GPS高程为正常高的算法,并与曲面拟合方法比较分析。经实例验证,在较大范围内,用神经网络方法转换GPS高程优于二次曲面拟合方法,所获得的正常高可满足工程生产的精度要求,具有一定的实用价值。

Research on GPS height conversion by BP neural network

For some current questions such as the large-scale uneven distributed points,the distortion effect on old elevation conversion,model error,and so on,an improved BP neural network method was presented for converting GPS Elevation algorithm in this paper,and the comparative analysis were made with surface fitting method.Test example result showed that the BP neural network method GPS elevation conversion is superior to conicoid fitting method in a larger degree,and the result can meet the accuracy requirement of work.