摘要
介绍了神经元网络BP算法的基本结构 ,提出了转换GPS高程的神经元网络模型———经改进的 5层BP网络结构 .该BP网络结构包括 :输入转换层、输入层、隐含层、输出层和输出转换层 .由于BP网络激活函数的数据限定区间为 [0 ,1],在工程应用中 ,增加设置输入转换层和输出转换层是必要的 .通过某工程实例 ,对 5层BP网络的具体模型结构进行了一些试验研究 ,如输入输出层的结构设计、隐含层最佳节点数的选取等 .最后得到了一些有工程实用价值的结论 .
The basic structure of BP algorithm of neural network is briefly introduced, and a neural network model-an advanced five-layer structure of BP network is proposed for GPS height transformation. The present neural network model consists of an input transformation layer, an input layer, an implicit layer, an output layer and an output transformation layer. For ordinary engineering application, the input transformation layer and output transformation layer are needed because the input and output of the Sigmoid standard active function f(x) are limited in the range of 0 to 1. Combined with a practical project, a series of experimental studies are made on the structure of the specific model, such as the structural design of the input and output layer, and the selection of the optimal node number for the implicit layer. Finally some conclusions of practical value are drawn.
出处
《河海大学学报（自然科学版）》
CAS
CSCD
北大核心
2001年第6期87-89,共3页
Journal of Hohai University(Natural Sciences)
