摘要
提出了自回归滑动平均(ARMA)模型的平方根超定递推辅助变量定价方法.该方法避免了计算逆协方差阵失去正定性的问题,改进了超定递推辅助变量定阶方法的数值稳定性.模拟计算与实验数据处理表明,该方法数值稳定性好,收敛速度快,抗噪能力强.是一种高性能的定阶方法。
A square-root overdetermined recursive instrumental vatiablealgorithm for ARMA models order estimation is presented.The methodavoids the problem of loss of positive-definiteness of the inverse covariance matrix and improves the numerical stability of the overdeterminedrecursive instrumental variable method for order estimation.the behaviorof the algorithm is illustrated by extensive numerical examples.Theresults of testing data processing and some simulation examples show thatthe method has good numerical stability, rapid convergence and strongnoise rejection.
出处
《湘潭大学自然科学学报》
CAS
CSCD
1991年第4期131-138,共8页
Natural Science Journal of Xiangtan University
关键词
ARMA模型
平方根法
定阶
autoregressive moving average models
square root method/overdetermined recursive instrumental variable method
square root overdetermined recursive instrumental variable method