| 基于主成分分析的BP神经网络长期预报模型 |
| |
| 引用本文： | 农吉夫,;黄文宁.基于主成分分析的BP神经网络长期预报模型[J].广西师院学报,2008(4):46-51. |
| |
| 作者姓名： | 农吉夫 ;黄文宁 |
| |
| 作者单位： | [1]广西民族大学数学与计算机科学学院,广西南宁530006; [2]河池学院数学系,广西宜州546300 |
| |
| 基金项目： | 国家自然科学基金项目（40675023）;广西民族大学青年科研基金项目（2007QN23） |
| |
| 摘 要： | MATLAB是进行神经网络系统设计及多元统计分析的有力工具．利用MATLAB6.5对月平均降水量的前期预报因子进行主成分分析，实现样本的最优压缩，从而降低样本的维数，建立起基于主成分分析的神经网络广西北部地区5月平均降水预测模型．计算结果表明，基于主成分分析的神经网络模型在预测中与多元回归模型相比有较好效果．
|
| 关 键 词： | 主成分分析 学习矩阵 BP神经网络 |
Long-term Prediction Model of BP Neural Networks Based on Principle Component Analysis |
| |
| Institution: | NONG Ji-fu, HUANG Wen-ning (1. College of Mathematics and Computer Science, Guangxi University for Nationalities, Nanning 530006; 2. Department of Mathematics, Hechi University, Yizhou 546300, China) |
| |
| Abstract: | MATLAB is a powerful instrument of designing a neural network system and multivariate statistical analysis. Principle component analysis was employed for the previous forecast factors of monthly mean precipitation by using MATLAB 6.5, and produce optimal compression of samples and reduce dimensions of samples. A neural networks forecast model with the monthly mean rainfall in May in the central part of Guangxi is developed based on principle component analysis. The calculating results show that the model of the neural networks based on Principle component analysis has better effect than the multivariable regression model in forecasting. |
| |
| Keywords: | principle component analysis learning matrix BP neural networks |
| 本文献已被 维普 等数据库收录！ |
