时变序列分析方法  

METHOD FOR TIME-VARYING SERIES ANALYSIS

作　　者：王治华[1] 傅惠民[1] 

机构地区：[1]北京航空航天大学小样本技术研究中心,北京100083

出　　处：《机械强度》2006年第3期353-357,共5页Journal of Mechanical Strength

基　　金：国防科技预研项目(413200204)资助。~~

摘　　要：对白噪声标准差随时间变化的时变序列模型进行研究,给出便于工程应用的模型形式。建立确定白噪声标准差、自回归系数和滑动平均系数函数形式的方法,并分别采用最小二乘法和极大似然法确定其中的待定参数。在此基础上建立时变序列预测公式及误差估计公式,给出其回归与时变自回归模型。The time-varying series model whose white-noise standard deviation varies with time is researched. A method for determining the ftmetion forms of white-noise standard deviation, autoregressive and moving average eoeflqeients is established. The least square method and the maximum likelihood method are used respectively to determine the parameters. The prediction formula and its error estimation are also established. Its regression and time-varying autoregression model is presented.

关 键 词：非平稳序列 时变序列 时变参数 预测 时变自回归 

分 类 号：O241.7[理学—计算数学] TB115[理学—数学]