多变量方程误差类系统的部分耦合迭代辨识方法

针对多变量方程误差滑动平均系统，利用最小二乘原理和迭代搜索原理，给出了增广随机梯度辨识方法、递推增广最小二乘辨识方法、梯度迭代辨识方法和最小二乘迭代辨识方法.针对多变量方程误差滑动平均系统和多变量方程误差自回归滑动平均系统，将多变量系统分解为一些子系统，利用耦合辨识概念，讨论了梯度迭代辨识方法、部分耦合（子系统）梯度迭代辨识方法、子系统最小二乘迭代方法和部分耦合子系统最小二乘迭代辨识方法.进一步结合数据滤波技术，研究了多变量方程误差自回归滑动平均系统的子系统梯度迭代辨识方法、部分耦合（子系统）梯度迭代辨识方法、部分耦合子系统最小二乘迭代辨识方法.文中给出了几个典型算法的计算步骤.

Partially coupled iterative identification methods for multivariable equation error type systems

For multivariable equation error moving average systems,this paper gives an extended stochastic gradient identification algorithm,a recursive extended least squares identification algorithm,a gradient based iterative(GI) identification algorithm and a least squares based iterative(LSI) identification algorithm,using the least squares principle and the iterative search principle.For multivariable equation error moving average systems and multivariable equation error autoregressive moving average systems,this paper uses the coupling identification concept and decomposes a multivariable system into several subsystems,derives the corresponding GI identification algorithm,partially coupled(subsystem) GI identification algorithm,subsystem LSI identification algorithm and partially coupled subsystem LSI identification algorithm.Furthermore,using the filtering technique,this paper studies the subsystem GI identification algorithm,partially coupled (subsystem) GI identification algorithm,partially coupled subsystem LSI identification algorithm for multivariable equation error autoregressive moving average systems.The computational steps for several typical identification algorithms are provided.