传递函数辨识(8):基于正弦响应的参数估计方法

Transfer Function Identification.Part H:Parameter Estimation Methods Based on the Sine Responses

针对不同极点惯性环节并联系统和不同极点惯性环节串联系统,利用多频组合正弦信号作为输入,通过测量系统的输出数据,基于二次优化和非线性优化技术,推导了估计传递函数参数的最小均方算法、随机梯度算法、多新息随机梯度算法、递推梯度算法等,并与梯度迭代算法、牛顿迭代算法相结合,提出了辨识传递函数参数的耦合递推—迭代辨识算法,文中的方法可以推广用于其他传递函数描述的动态系统参数辨识,如具有共轭极点、重极点传递函数参数的辨识以及任意非线性函数的参数估计。

For parellel systems and series systems which consist of the inertial links with distinct poles,using composite multi-frequency sinusoidal signals as the input of the systems and measuring the output data of the systems,we derive the least mean square algorithm,stochastic gradient(SG)algorithm,multi-innovation SG algorithm,and recursive gradient algorithm for etimating the parameters of the transfer functions.Furthermore,combining the gradient-based iterative algorithm and the Newton iterative algorithm,we propose the corresponding coupled recursive-iterative identificative algorithms for identifying transfer functions.The proposed methods can be extended for the parameter identification of dynamical systems described by other transfer functions with conjugate poles and with same poles and nonlinear functions.