摘要
基于对神经元模型、网络拓扑结构和学习算法3要素的描述,系统地剖析了一个BP3-2-1结构的模型模拟非线性函数的一般过程。基于杉木人工林密度实验数据,以初始密度、立地指数和林龄为输入,林分平均胸径为输出,利用MATLAB(R2010b)神经网络工具箱创建、训练BP模型,模型的均方误差Ems=0.01,实测值和预测值之间的相关系数R=0.977。
The aim of this paper is to systematically reveal the principle of BP model with 3-2-1 structure according to the introduction of neuron model, topology and back-propagation algorithm for supervised learning. A BP model is set up in the MATLAB （R2010b） environment to simulate the mean diameter at breast height of a Cunninghamia lanceolata plantation, using initial density, site index and stand age as input parameters. The mean square error of the model is 0.01 and the correlation coefficient between the independent variable and the dependent variable is 0. 977.
出处
《东北林业大学学报》
CAS
CSCD
北大核心
2011年第8期116-119,共4页
Journal of Northeast Forestry University
关键词
BP神经元模型
网络拓扑结构
学习算法
胸径模拟
BP neuron model
Network topologies
Learning algorithms
Simulation of diameter at breast height
