| BP神经网络隐层单元数的确定方法及实例 |
| |
| 引用本文： | 严鸿,管燕萍.BP神经网络隐层单元数的确定方法及实例[J].控制工程,2009(Z2). |
| |
| 作者姓名： | 严鸿 管燕萍 |
| |
| 作者单位： | 上海交通大学自动化系;中兴通讯股份有限公司; |
| |
| 基金项目： | 国家863基金资助项目(2007AA041403,2006AA040308);;上海启明星计划基金资助项目(07QA14030);;上海经委创新基金资助项目(07XI-041) |
| |
| 摘 要： | 针对BP神经网络隐层单元数不易确定的问题,提出一种在传统的经验公式基础上快速确定隐层单元数的方法。该方法首先借助经验公式确定隐层单元数的取值范围,然后将其扩大,在这个扩大的范围内寻找最优值。以BP神经网络预测交通流量为例,解释说明了具体的步骤,以及网络模型的隐层结构对模型仿真精度的影响。结果表明,采用该方法可快速决定隐层单元数,在实例中采用16个隐层单元数为最佳。
|
| 关 键 词： | BP神经网络 网络结构 隐层单元数 车流量预测 |
Method to Determine the Quantity of Internal Nodes of Back Propagation Neural Networks and Its Demonstration |
| |
| YAN Hong,GUAN Yan-ping.Method to Determine the Quantity of Internal Nodes of Back Propagation Neural Networks and Its Demonstration[J].Control Engineering of China,2009(Z2). |
| |
| Authors: | YAN Hong GUAN Yan-ping |
| |
| Affiliation: | 1.Automation Department;Shanghai Jiaotong University;Shanghai 200240;China;2.ZTE Corporation;Shanghai 201203;China |
| |
| Abstract: | A new scheme is proposed based on the experiential formula to determine the quantity of internal nodes of a ANN.The method uses the experiential formula to determine the range,to expand it and then find the best one.Traffic flow prediction by using BP neural network is shown as an example to show the effectiveness of the method to determine the quantity of internal nodes and explain the effect of the choices.As a result,16 is the best choice as the internal nodes. |
| |
| Keywords: | BP neural network structure of the networks number of internal nodes traffic flow prediction |
| 本文献已被 CNKI 等数据库收录！ |
