摘要
本文对多层递阶预报方法进行了深入的研究,提出了一种初值选取方法,并选择AR模型作为预报模型,采用高效的Houscholder变换来确定模型阶次,从而使得这种预报方法简便而实用,预报精度亦有进一步提高。本文还提出一种季节性时间序列预报的新方法,它具有简单、易行,且精度高的特点。本文将多层递阶预报方法成功地应用于冷库自动控制系统,实现了对温度的实时预报,取得了满意的效果。
In this paper, a new method to choose initial value is developed for the method of multi-level recursive prediction, where AR model is used as prediction model. The householder matriex has been used to identify the order of AR model. The paper presents a new prediction method for seasonal time series.At last,the recursive prediction method is used to predict the temperature of cool storage real-timely.
出处
《控制与决策》
EI
CSCD
北大核心
1991年第6期428-433,共6页
Control and Decision
关键词
递阶预报
时间序列
应用
multi-level recursive prediction, seasonal time series
