中文摘要：

介绍小波神经网络的基本结构及学习算法,并应用于GPS大地高转换为正常高。结合实际工程数据,与BP神经网络作比较分析,因小波网络较强的非线性使得它泛化性能更好,收敛速度更快,经实例论证,在同等条件下,小波神经网络方法用于GPS高程转换的精度优于BP神经网络,且其精度可满足常规工程需要,具有一定实用价值。

英文摘要：

In this study, the basic structure of the wavelet neu- ral network and learning algorithm was described, and applied to the transition of GPS Height to Normal Height. With actual project data, wavelet network generalization performances bet- ter and faster convergence rate due to its strong non-linear com- pared with BP neural network under the same conditions dem- onstrated by the instance, and it has higher accuracy than BP neural network in GPS height conversion, which can meet in the traditional engineers＇ accuracy, which is proved to have certain practical value.