输入输出层选取对GPS高程转换影响  

Influence of Input and Output Layer Selection on GPS Elevation Transformation

作　　者：王伟 吴海兵 Wei Wang;Haibing Wu(Shanghai Investigation Design and Research Institute Co., Ltd., Shanghai 200434, China)

机构地区：[1]上海勘测设计研究院有限公司,中国上海200434

出　　处：《工程技术与管理（新加坡）》2019年第6期31-34,共4页Engineering Technology & Management

摘　　要：利用神经网络转换GPS 高程来精化区域似大地水准面需要选择合理的输入输出层,本文论述了如何设计输入输出层,并在基于不同的输入输出层基础上运用标准BP 神经网络模型、GRNN 神经网络模型和RBF 神经网络模型进行运算,将转换结果进行对比,得出不同的输入输出层会对GPS 高程转换结果产生一定的影响,并对实际应用中输入输出层的选择提出合理的意见.Selecting a reasonable input and output layer is necessary in GPS elevation transformation based on neural network in regional Quasi- geoid determination. Then it discusses how to design the input and output layer and uses the standard BP neural network model, GRNN neural network and RBF neural network to calculate the transformation and compare the results based on different input and output layer .It was verified that different input and output layer could generate some influences on GPS elevation transformation, and finally the authors give some suggestion on the selection of input and output layer.

关 键 词：输入输出层 神经网络 GPS高程转换 

分 类 号：P228.4[天文地球—大地测量学与测量工程]