降水时间序列的聚类分析和预测

王永县;詹一辉;张少竑

系统工程理论与实践 ›› 1994, Vol. 14 ›› Issue (11) : 67-71.

PDF(327 KB)
PDF(327 KB)
系统工程理论与实践 ›› 1994, Vol. 14 ›› Issue (11) : 67-71. DOI: 10.12011/1000-6788(1994)11-67
论文

降水时间序列的聚类分析和预测

    王永县, 詹一辉, 张少竑
作者信息 +

Clustering Analysis and Prediction for Precipitation Time Series

    Wang Yongxian, Zhan Yihui, Zhang Shaohong
Author information +
文章历史 +

摘要

本文阐述了对160个中国大陆降雨序列进行同步预测的完整方法。预测模型是用多元分析和随机序列等方法综合建立的, 预测结果较好。文中重点研究能用个人计算机可方便地同时预测这160个点的方法。因此, 首先采用聚类方法将160个点的降水序列分成若干个局部特征相似的几个子类。采用该法的分类结果与国内大多数着名专家的分类结果相似。采用数据压缩法求出每个子类周期性特征的主成分序列, 计算中在精度损失不大情况下尽量压缩计算量。该模型预测一年后的降雨量, 获得令人满意的结果。

Abstract

In this paper we presents an integrated approach to simultaneous forecasting for the 160 precipitation series distributed on mainland China.A forecasting model has been established by combining powerfull methods in multivaraiate analysis and random time series analysis.The forecasting results are good,Our problem is one of developing a computationally tractable model on personal computers for the simultaneous forecasting of the 160 precipitation series,Cluster-ing method is applied to classify the 160 precipitation series into locally similar subgroups.The classification so obtained resembles those obtained by most famous Chinese scientists.A princi-pal component series reserving the periodic characteristics of each subgroup can then be extract-ed by data suppression method,allowing a considerable reduction of computational burden with relatively small loss of prediction precision. One year ahead prediction by our model shows promising results.

关键词

聚类分析 / 降水预测 / 主导序列

Key words

Clustering Analysis / Precipitation Prediction / Principal Series

引用本文

导出引用
王永县 , 詹一辉 , 张少竑. 降水时间序列的聚类分析和预测. 系统工程理论与实践, 1994, 14(11): 67-71 https://doi.org/10.12011/1000-6788(1994)11-67
Wang Yongxian , Zhan Yihui , Zhang Shaohong. Clustering Analysis and Prediction for Precipitation Time Series. Systems Engineering - Theory & Practice, 1994, 14(11): 67-71 https://doi.org/10.12011/1000-6788(1994)11-67
中图分类号: O21   
PDF(327 KB)

722

Accesses

0

Citation

Detail

段落导航
相关文章

/