摘要
即使对于线性时不变系统,脉冲响应也是系统参数的高度非线性函数。利用系统的脉冲响应观测数据和非线性优化方法,论文提出了一阶系统、二阶系统、高阶单极点系统传递函数的最小均方参数估计方法、随机梯度参数估计方法、多新息随机梯度参数估计方法等,以及高阶单极点系统的递推梯度参数估计算法和牛顿递推参数估计方法等。提出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、重极点高阶传递函数参数的辨识。
Even for linear time-invariant systems,the impulse response is the highly nonlinear functions of system parameters.Using the system′s impulse response observation data and nonlinear optimization methods,this paper presents least mean square parameter estimation methods,stochastic gradient parameter estimation methods,multi-innovation stochastic gradient parameter estimation methods for the transfer functions of first-order systems,second-order systems and higher-order systems with distinct poles,and recursive gradient parameter estimation methods and Newton recursive parameter estimation methods of higherorder systems with distinct poles.The proposed methods can be extended for the parameter identification of dynamical systems described by other transfer functions,such as high-order transfer function with conjugate poles and with same poles.
作者
丁锋
徐玲
刘喜梅
DING Feng;XU Ling;LIU Xime(College of Automation and Electronic Engineering,Qingdao University of Science and Technology,Qingdao 266042,China;School of Internet of Things Engineering,Jiangnan University,Wuxi 214122,China)
出处
《青岛科技大学学报（自然科学版）》
CAS
2018年第4期1-15,共15页
Journal of Qingdao University of Science and Technology:Natural Science Edition
基金
国家自然科学基金项目(61472195)
关键词
传递函数
参数估计
系统辨识
脉冲响应
梯度搜索
牛顿搜索
transfer function
parameter estimation
system identification
impulse response
gradient search
Newton search
