【作者】 刘丽兰； 刘宏昭；

【Author】 LIU Li-lan, LIU Hong-zhao (Department of mechanical and instrumental engineering, Xi’an University of Technology, Xi’an, Shaanxi, 710048)

【机构】 西安理工大学机械与精密仪器工程学院；

【摘要】 本文研究了对观测随机序列建立时序模型的初步识别问题,包括模型形式的确定和模型阶次的选择．介绍了一套完整的模型识别的方案．首先利用样本自相关与偏自相关函数的截尾特性,判断时序模型的形式,并可初步选择自回归模型或滑动平均模型的阶次．然后再根据模型定阶准则最终确定模型的阶次．数字算例结果表明了该套方法的可行性．更多还原

【Abstract】 In this paper, the identification problem of time series models building for observed random data is discussed, including the determination of model type and model order. A set of complete identifying program is introduced. First, the truncated characteristics of the autocorrelation and the partial autocorrelation functions are used to determine the model type. Simultaneously, the order of the autoregressive model or the moving average model can be initially selected. Then, according to the order selection criterions, the model order can be eventually determined. The result of the numerical example shows the feasibility of the method.更多还原