动态加权预测方法设计及在降水预测中的应用

Design and application of dynamic weighted forecasting method to precipitation forecast

在动态数据预测中,为克服传统预测方法中所造成的较大误差,本文通过动态预测理论与移动平均预测理论相结合,设计了一种加权移动平均预测方法,用来预测时间序列数据.在原有的移动平均预测基础上构造一个动态加权参数.通过对相对误差的比较,证实了该方法预测效果优于移动平均预测.并利用该方法对乌鲁木齐地区30年平均降水量进行预测研究.并与动态预测方法相比较,预测效果相对误差较小.

In traditional prediction methods,we have to overcome the errors when predict for the dynamic data. In this paper,the dynamic prediction theory and moving average forecasting has been combined. We had designed a kind of weighted moving average forecasting method to predict the data of time series. Dynamic weighting parameter has been constructed in the original moving average forecast. The prediction effect is better than moving average forecast when compared the relative error. At last,we predicted the Urumqi prediction of precipitation for 30 years with this method and compared with the dynamic prediction method,the effect of the prediction of relative error is smaller.