多变量ARMAX模型结构和参数的辨识

Structure Identification and Parameter Estimation of Multivariate ARMAX

本文讨论了多变量ARMAX模型结构和参数的辨识问题。根据[1]的论述,采用BIC准则判定模型的阶;并用递推的极大似然法辨识参数。根据估计参数的渐近正态性进一步确定模型的子阶和时滞。由于采用了BIC准则,避免了F检验定阶的主观性。因此本方法比文献[2]、[3]中的方法优越。仿真算例证明了本方法的有效性。

This paper discusses the problem of structure identification and parameter estimation of multivariate ARMAX.BIC criterion is used to determine the order of the model according to [1].Recursive maximum likelihood is used to estimate the parameters.The suborder and delay of the model are determined according to the asymptotical Gaussian distribution of the estimated parameter.Because of the application of BIC criterion,the subjectivity of F criterion has been avoided.So, the method is superior to the method introduced in [2],[3].The effectiveness of the proposed method is illustrated by a simulation example.