针对一类有限时间区间上具有可重复性的BIBO稳定的一阶线性时变系统,将模型参考自适应辨识方法与迭代学习相结合,提出了模型参考自适应迭代学习的参数辨识算法。利用模型参考自适应辨识方法得到时变系统参数辨识结构,针对系统可重复的特点,基于Lyapunov方法得到时变参数的迭代学习律。该算法可以辨识快时变的参数,而不需要参数时变结构的信息,并可保证参数估计误差和模型输出误差有界,且沿迭代轴逐点收敛。分析了参数收敛到真值的条件,系统仿真验证了辨识算法的有效性。
A model reference adaptive iterative learning identification algorithm is proposed for a class of first order linear time-varying systems which are BIBO stable and repeatable in finite time interval. The identification structure of the time-varying system is obtained by combining the model reference adaptive identification approach with iterative learning. Based on the repeatability of the system, the iterative learning law for time-varying parameters is given by means of Lyapunov method. Rapidly time-varying parameters can be identified without information about the time-varying structure of unknown parameters. The boundedness and pointwise convergence along the iteration horizon can be guaranteed for the parameter errors and output tracking errors. The condition that the parameters converge to the true parameters is analyzed. The effectiveness of the algorithm is demonstrated by simulations.