多元系统耦合带遗忘因子有限数据窗递推最小二乘辨识方法-控制与决策.pdfVIP

第31 卷 第10 期 控 制 与 决 策 2016 年 10 月 Vol. 31 No. 10 Control and Decision Oct. 2016 文章编号: 1001-0920 (2016) 10-1765-07 DOI: 10.13195/j.kzyjc.2015.1214 多元系统耦合带遗忘因子有限数据窗 递推最小二乘辨识方法 时振伟, 纪志成, 王 艳 (江南大学物联网工程学院，江苏无锡214122) 摘 要: 针对多元线性或非线性回归系统, 将耦合辨识思想与带遗忘因子有限数据窗辨识理论相结合, 提出一种耦 合带遗忘因子有限数据窗递推最小二乘辨识算法. 该算法每次递推计算时既不涉及矩阵求逆运算, 又可以克服数据 饱和现象, 因此, 该算法不仅计算效率高, 而且可以快速地跟踪时变参数, 获得精确的参数估计. 通过辨识基于多元模 型的永磁同步电机参数的实例, 验证了所提出算法的有效性和实用性. 关键词: 多元系统；有限数据窗；最小二乘；参数辨识；永磁同步电机 中图分类号: TP273 文献标志码: A Coupled finite-data-window RLS identification approache with forgetting factors for multi-variate systems SHI Zhen-wei, JI Zhi-cheng, WANG Yan (School of Internet of Things Engineering，Jiangnan University ，Wuxi 214122 ，China. Correspondent ：JI Zhi-cheng, E-mail ：zcji@jiangnan.edu.cn) Abstract: For multi-variable linear/nonlinear regression systems, this paper combines the coupling identification idea with the finite-data-window with forgetting factors theory to present a coupled finite data-window RLS algorithm with forgetting factors. Because the proposed algorithm can avoid the matrix inversion at each recursive calculation and can overcome the data saturation phenomenon, it not only has a highly efficient computation, but also tracks effectively the time-varying parameter and obtains accurate parameter estimates. The application example of parameter identification of the permanent magnet synchronous machine based on the multi-variable model shows the effectiveness and feasibility of the proposed algorithm. Keywords: multi-variate systems ；finite-data-window ；RLS ；parameter identification ；permanent magnet synchro