摘要
建立一种考虑了影响系数的优势周期均生函数主分量时序须测模型。该方法对时间序列的均生函数进行主分量分析，用主分量作自变量。根据变量后期值受前期值的影响引入影响系数Ｋ（ｔ），并应用方差分析法确定Ｋ（ｔ）的优势周期，进而求出预测时刻的Ｋ（Ｎ＋１），由变量的前期值和Ｋ（Ｎ＋１）就能计写出变量的预测值（Ｎ＋１），该方法用于四川省２０个地区丰季和全年降水预测，并将试报结果与均生函数建模法试报结果进行了比较，效果尚好。
In this paper,we present a forecasting model of temporal sequence based on prineipal component of mean generating function taking into consideration the superiurity period of influence coefficient.The model is formulated through analysing the principal component of mean generating function fortemporal sequence. Based on the definition of influence coefficlent K（t）in which the earlier stagevalues affect the later stage ones and the superiority peried of K（t）determined with variance analysis，the influence coefficient of time point in the future can be calculated．Finally，the predicting results ofvariable is derived on the basis of the principal component of earlier stage variable and K（N＋1）．Themethod is used in forecasting medelling of two kinds of precipitation of 2 0 regions in Siehuan，respectively. The comparison between test forecast results and that of mean generating function showsthat the efrcet of this model is much better.
基金
国家自然科学基金资助课题
关键词
优势周期
主分量
均生函数
预测模型
降水预报
：Superiority period
Extrapolation matrix
Principal component
Mean generating function；Prediction model.
