| 支持向量机方法在郑州冬季气温趋势预测中试应用 |
| |
| 引用本文： | 李素萍,常军,朱业玉,焦建丽,陈静.支持向量机方法在郑州冬季气温趋势预测中试应用[J].河南气象,2006(1):15-16. |
| |
| 作者姓名： | 李素萍 常军 朱业玉 焦建丽 陈静 |
| |
| 作者单位： | 河南省专业气象台 河南郑州450003(李素萍)，河南省气候中心 河南郑州450003(常军，朱业玉，焦建丽，陈静) |
| |
| 摘 要： | 将气候预测中常用的74项环流特征量进行归一化处理后,与郑州市冬季气温进行相关普查,利用SVM(SupportVectorM ach ine)两类分类方法,同时考虑气温的年代际变化,建立郑州冬季温度距平趋势预测推理模型,并对因子个数多少和年代际变化对预测模型的影响进行了试验。试验结果表明:用25个和15个因子分别建模时,产生最优模型时样本平均Ts评分均为56%,但后者预报准确率为75%,较前者提高了10%。用20世纪50年代和60年代做试验集,效果较好,产生最优模型时的样本Ts评分和预报准确率较高;用90年代做试验集,效果较差。
|
| 关 键 词： | 支持向量机 训练集 实验集 检验集 |
| 文章编号： | 1004-6372（2006）01-0015-02 |
| 收稿时间： | 2005-09-12 |
Application of Support Vector Machine Method in Winter Temperature Forecast of Zhengzhou |
| |
| LI Su -ping, CHANG Jun , ZHU Ye -yu , JIAO Jian -li , CHEN Jing.Application of Support Vector Machine Method in Winter Temperature Forecast of Zhengzhou[J].Meteorology Journal of Henan,2006(1):15-16. |
| |
| Authors: | LI Su -ping CHANG Jun ZHU Ye -yu JIAO Jian -li CHEN Jing |
| |
| Institution: | 1. Special Meteorological Observatory of Henan Province, Zhengzhou 450003, China; 2. Henan Climate Center, Zhengzhou 450003, China |
| |
| Abstract: | Normalizing the 74 kinds of characteristic quantity on circulation which are broadly used in weather forecast,the author proceeds a widespread relation check between these characteristic quantity and the winter temperature of Zhengzhou.Considering the decadal temperature variability,he builds a reasoning model on Zhengzhou's wintertime temperature anomalous trend prediction by making use of the two-classification method of SVM.Besides,he makes a test about how much the number of the factors and the decadal variability influence upon the forecast model. |
| |
| Keywords: | Support Vector Machine(SVM) Training Data Testing Data Checking Data |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
