摘要
本文利用时间序列建立了一个趋势回归模型、虚拟参数的季节模型和时间序列分析的统计预测模型.该模型能够很好地分离出时间序列中的趋势成分,且能够很好地刻画它在各个年度周期内部光滑的季节成分.该文将地下水位统计数据进行了消除趋势成分及消除季节影响的处理,再将新得到的数据进行平稳化、零均值化处理,并利用时间序列的自相关函数、偏自相关函数的性质,最终确认了适当的时间序列模型.以北京市1990-1994年共60个月份的地下水位统计数据为实例进行分析,确定此模型为ARMA（1,4）模型.通过对数据的计算,可预测出北京市1995年地下水位值.
In this paper it builds up a forecasting model, which is combined with the ordinary linear regression （ least square estimate） model, non-parametric season estimation and time series analysis. The proposed model can well separate the trend component of the data and depict the smooth season component of every month. It eliminates trend component and season component＇s influences in the data, dealing into smooth and zero expectative data with the new data and determines appropriate model. By analyzing the data of Beijing＇s ground water level data from 1990 to 1994, it defines as the ARMA（ 1,4） model from AC and PACt characters and the selection standards. Through computing the data, it can forecast Beijing city＇s ground water level in 1995. It indicates the result that the ideas have the more precision through RSS.
出处
《天津理工大学学报》
2008年第2期8-10,共3页
Journal of Tianjin University of Technology
基金
国家自然科学基金(1047207750375107)
