对NWS中资源性能预报算法的一个改进  

Improvement to Forecasting Algorithms for Performance of Resource in NWS

作　　者：张海洋[1] 鞠九滨[1] 胡亮[1] 

机构地区：[1]吉林大学计算机科学与技术学院,长春130012

出　　处：《吉林大学学报（理学版）》2005年第2期157-161,共5页Journal of Jilin University:Science Edition

基　　金：国家自然科学基金(批准号:60473099);吉林省杰出青年基金(批准号:20040119)

摘　　要：提出一种新的预报算法———动态指数平滑算法,该算法是对传统的指数平滑算法的改进.为了使指数平滑算法能够更好地适应NWS系统对精确度和系统开销的要求,增加了对参数的调整.但这种调整不同于已有的参数调整算法,并且在时间复杂性、空间复杂性和预报准确性方面均得到了有效的提高.通过实验和比较,本算法更适用于NWS中的时间序列以及有着相似统计学特性的时间序列.In this paper a new forecasting method, the dynamic exponential smoothing algorithm is implemented, which is derived form the exponential smoothing algorithm. The parameters in the new method can be regulated dynamically, which will make the method more effective in the NWS system. This new forecasting method is different from the existing exponential smoothing algorithm, which can also regulate parameters, and it may be useful in the time series of NWS and other time series with the same statistic characteristics. Compared with the existing method in NWS, this method is better in terms of time complexity, space complexity and forecasting accuracy.

关 键 词：网络气象服务系统 CPU可用性 网络带宽 时间序列分析 指数平滑算法 动态指 数平滑算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]