H∞辨识的多项式插值算法及Worst－Case误差

A Polynomial Interpolation Algorithm and the Worst Case Error in H ∞ System Identification

就单输入、单输出、离散、时不变因果系统讨论了面向鲁棒控制的Ｈ∞系统辨识问题．首先提出了用线性规划检验先验假设和后验数据是否相容的判据，然后给出了基于线性规划的多项式插值算法，该算法计算量少、所得模型简单，最后估计了该算法的Ｗｏｒｓｔ－Ｃａｓｅ误差。

The robust control oriented H ∞ system identification problem is studied. The system to be identified is assumed to be a single input, single output, discretized or time invariant causal one. A criterion for detecting if it is consistent between the a priori suppositions and a posterior data is proposed using linear programming. A polynomial interpolation algorithm based on linear optimization theory is suggested, which requires short computing time and the model developed is simple. The worst case identification error associated with the algorithm is evaluated and the convergence of the algorithm is proved.