滤波辨识(10):多变量Box-Jenkins系统的滤波辅助模型递阶广义增广参数辨识

Filtering Identification.Part J:Filtering-Based Auxiliary Model Hierarchical Generalized Extended Parameter Identification for Multivariable Box-Jenkins Systems

针对多变量Box-Jenkins模型,即多变量输出误差自回归滑动平均(M-OEARMA)系统,利用滤波辨识理念和辅助模型辨识思想,研究和提出了滤波辅助模型递阶广义增广随机梯度辨识方法、滤波辅助模型递阶多新息广义增广随机梯度辨识方法、滤波辅助模型递阶广义增广递推梯度辨识方法、滤波辅助模型递阶多新息广义增广递推梯度辨识方法、滤波辅助模型递阶广义增广最小二乘辨识方法、滤波辅助模型递阶多新息广义增广最小二乘辨识方法。这些滤波辅助模型递阶广义增广辨识方法可以推广到其他有色噪声干扰下的线性和非线性多变量随机系统中。

For multivariable output-error autoregressive moving average(M-OEARMA)models,which are also called multivariable Box-Jenkins models,this paper investigates and proposes filtered auxiliary model hierarchical generalized extended stochastic gradient identification methods,filtered auxiliary model hierarchical multi-innovation generalized extended stochastic gradient identification methods,filtered auxiliary model hierarchical generalized extended recursive gradient identification methods,filtered auxiliary model hierarchical multi-innovation generalized extended recursive gradient identification methods,filtered aux-iliary model hierarchical generalized extended least squares identification methods,and fil-tered auxiliary model hierarchical multi-innovation generalized extended least squares identi-fication methods by using the filtering identification idea and the auxiliary identification idea from available input-output data.These filtered auxiliary model hierarchical generalized ex-tended identification methods can be extended to other linear and nonlinear multivariable sto-chastic systems with colored noises.