几种基于神经网络的GPS高程拟合方法比较

摘要

为提高GPS大地高向正常高转换的精度，本文对目前研究较广泛的 BP神经网络方法、遗传神经网络方法和退火神经网络方法用于 GPS高程拟合的特点和拟合精度进行比较分析，为使用这些方法进行GPS高程拟合提供了参考。%Characteristics and precision of BP neural network,genetic neural network and annealing neural network applied in GPS height fitting were hereby compared and ana-lyzed to improve the precision of transforming GPS geodetic height into normal height,and improvement approaches were proposed with respect to all problems in GPS height fitting method.