多个未知时延的miso系统的递推辨识.pdfVIP

第25 卷 第1 期 2010 年1 月 Vol.25 No. 1 Control and Decision Jan. 2010 : 10012 0920(2010) 012 00932 06 MISO , , ( 南京航空航天大学 自动化学院, 南京 210016) : 对于未知时延的多输入单输出( MISO) 系统, 借助分离性原理, 推导出迭代的可分离的 线性最小二乘 ( SNLS)辨识方法. 为降低收敛于局部最小的可能性, 利用全局优化理论, 推导了全局可分离的 线性最小二乘 ( GSNLS) 辨识方法;为消除强观测噪声所引起的参数估计的偏差, 将GSNLS 方法调整为一新颖的全局可分离的 线性多新息递推最小二乘( GSNMIRLS) 辨识方法, 仿真实验验证了算法的有效性. : 递推辨识; 时延; 线性最小二乘法; 多新息; 全局优化 : T 273 : A Recursive identification of MISO systems with multiple unknown time delays WAN G J ian2hong , WAN G Dao2bo, WAN G Zhi2sheng ( College of Automation Engineer, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China. Correspondent: WANG Jian2hong, E2mail: wangjianhong@nuaa. ) Abstract: For MISO systems with multiple unknown t ime delays, an iterative separable nonlinear least2 squares ( SNLS) identification method is proposed by means of the separable principle. Then a global separable nonlinear least2 squares ( GSNLS) identification method which est imates the time delays and transfer function parameters separably is derived to reduce the possibility of convergence t o a local minimum by using the global opt imization theory. Furthermore, the GSNLS method is modified to a novel global separable nonlinear mult 2i innovation recursive least2 squares identificat ion ( GSNMIRLS) method to eliminate the biases of the estimates in the presence of high measurement noise. The simulation results show the theoretical results. Key words: Recursive identificat ion; Time delay; Nonlinear least2 squares method; Multi2innovation; Global optimization 1 引 言 法(SNLS) , 借助随机全局优化方法来表示收