GPS高程拟合中神经网络法的基本思想

The Application of Neural Network in GPS Height Fitting

高程问题一直是困扰GPS的问题之一,而不同的GPS高程拟合方法都有其适用条件,并且精度不等。本文在对人工神经网络的基本原理、神经元模型、网络结构、数据结构和训练方式等研究的基础上,给出一种新的算法(新BP算法)。以“阜新控制网改造工程”作为一个具体实例,使用Matlab语言来完成GPS高程的拟合,并与其他方法作了比较,最后给出有益的结论。

The problem of height is always one of the problems which confused GPS. And different GPS height fitting method has its own application conditions. This paper studied the basic principle of neural network, neural cell model, network structure, data structure and training method, and so on. Then it put forward a new algorithm （new BP algorithm）. Taking Control Network Reconstruction Project of Fuxin city as an example, it accomplished the GPS height fitting using Matlab Language and compared with other methods, then presented useful conclusions.