基于多维时间序列挖掘的降雨天气模型研究

为研究降雨天气中降雨量和相关气象要素的关系,找出降雨前后相关气象要素的变化规律,提出了多维时间序列数据挖掘模型.该模型首先对气象要素时间序列进行维度选择预处理,剔除不相关及冗余维度,然后运用提出的极值斜率分段线性拟合法对时间序列进行分段、数据压缩及特征值提取,最后使用k-means聚类算法对处理后的多维序列进行符号化,利用规则提取得到降雨天气模型.实验结果表明了该模型具有较好的实用价值.

Research of rainfall weather model based on multi-dimensional time series data mining

To research the relationship between rainfall and related meteorological factors and search for regularity of variety for those factors before and after rainfall, a multi-dimensional time series data mining model is presented. Firstly, in order to eliminate irrelevant redundant dimensions,the model performs dimensions selection for time series of meteorological factors.Then,a new extremum slope piecewise linear fitting method is introduced to implement multi-dimensional meteorological time series segmentation,data compression and feature value extraction.Finally,k-means clustering method is used to symbolize processed multi-dimensional series,and the rainfall model is extracted by rules extraction.The results of experiment show this model has agreat practicability.