非均匀周期采样系统的递阶最小二乘辨识方法  

Hierarchical least squares identification for periodically non-uniformly sampled systems

作　　者：刘艳君[1] 丁锋[1] 

机构地区：[1]江南大学控制科学与工程研究中心,江苏无锡214122

出　　处：《控制与决策》2011年第3期453-456,463,共5页Control and Decision

基　　金：国家自然科学基金项目(60973043).

摘　　要：针对非均匀周期采样系统,通过状态空间模型离散化方法得到其输入输出表达形式.鉴于参数化后得到的辨识模型同时包含1个参数向量和1个参数矩阵,利用递阶辨识原理,将辨识模型分解为分别含有参数向量和参数矩阵的2个虚拟子系统;考虑到系统的因果约束问题,将包含参数矩阵的子系统分解为子子系统进行辨识,从而提出这类非均匀采样系统的递阶最小二乘辨识方法.仿真例子表明该算法是有效的.An input-output representation is obtained for the periodically non-uniformly sampled systems by using the discretization technique of the state space models. According to the hierarchical identification principle, the obtained identification model is decomposed into two fictitious subsystems, one containing the parameter vector and the other containing the parameter matrix. Taking into account the causality constraints, the subsystem with the parameter matrix is also decomposed into several sub-subsystems, and the hierarchical least squares identification method is developed for the non-uniformly sampled systems. The simulation results show the effectiveness of the proposed algorithm.

关 键 词：递阶辨识 参数估计 非均匀采样 多率系统 状态空间模型 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]