论文部分内容阅读
鲁棒控制的系统辨识问题是控制领域一个十分活跃的研究方向。该文综述了具有正规互质因子扰动模型集的辨识与检验的结果。在闭环实验数据已知时 ,给出了正规互质因子频域响应极大似然估计的解析表达式 ,介绍了这一估计的统计特性和最优参数选择。讨论了基于受约束曲线拟合的正规互质因子参数辨识问题 ,给出了其解析解。最后介绍了模型集与控制器和闭环实验数据相匹配的一个充分必要条件。这一条件可表示为一个线性约束和一个复数向量的欧几里德范数。根据这一条件 ,讨论了随机框架下的模型集检验问题 ,给出了采样非伪概率的计算方法
The problem of system identification with robust control is a very active research field in control field. This paper reviews the results of identification and verification of set of perturbations with regular coprime factors. When the closed-loop experimental data are known, the analytical expression of the maximum likelihood estimation of the frequency response of normal coprime factor is given. The statistical properties of this estimate and the choice of optimal parameters are introduced. Discusses the parameter identification of normal coprime factor based on constrained curve fitting, and gives its analytical solution. Finally, a sufficient and necessary condition that the model set matches the controller and the closed-loop experimental data is introduced. This condition can be expressed as a Euclidean norm of a linear constraint and a complex vector. According to this condition, the problem of model test in random frame is discussed, and the calculation method of sampling non-pseudo-probability is given