摘要
线性时间序列模型谱密度的计算可以直接由定义获得,而非线性时间序列模型谱密度的计算目前还没有一般的理论.已有研究者将AR模型推广到MAR(混合自回归)模型,并且讨论了该模型的参数估计及模型选择问题.作者利用全期望公式及差分方程理论研究了混合自回归时间序列模型的谱分析,导出了自协方差函数的递推公式,给出计算谱密度的算法,并对一些常见的特殊情形给出了谱密度的具体表达式.
Spectual density computation of linear time series models can be achieved by definition. However, spectual density computation of nonlinear time series models have no general theory. Some authors generalized AR model to MAR model, and discussed parameter estimate and model selection problems of the model. This paper researched the spectral analysis of mixture autoregressive time series model by Law of total expectation and theory of difference equation. We first derived a recursive formula for the auto-covariance function, and then gave an algorithm for computing the spectral density. Some closed formula were given for some particular cases.
出处
《安徽大学学报(自然科学版)》
CAS
北大核心
2012年第5期31-36,共6页
Journal of Anhui University(Natural Science Edition)
基金
国家自然科学基金资助项目(10726072)
关键词
混合自回归模型
自协方差函数
谱分析
谱分布函数
谱密度
mixture autoregressive model
auto-covariance function
spectral analysis
spectral distribution function
spectral density.