上海交通大学学报(自然版) ›› 2014, Vol. 48 ›› Issue (07): 982-986.

• 自动化技术、计算机技术 • 上一篇    下一篇

慢时变线性模型参数辨识递推算法及收敛性分析

曹鹏飞,罗雄麟
  

  1. (中国石油大学 自动化研究所, 北京  102249)
     
  • 收稿日期:2013-07-03 出版日期:2014-07-28 发布日期:2014-07-28
  • 基金资助:

    国家重点基础研究发展规划(973)项目(2012CB720500)资助

Recursive Identification Algorithm and Its Convergence Analysis for Slow Time-Varying Linear Model

CAO Pengfei,LUO Xionglin
  

  1. (Department of Automation, China University of Petroleum, Beijing 102249, China)
  • Received:2013-07-03 Online:2014-07-28 Published:2014-07-28

摘要:

针对慢时变线性模型,给出辨识的递推算法,并证明该算法能够保证参数收敛在一个有界空间区域,该区域包含参数真值集合;若工作点不发生变化,合理的收敛因子保证参数收敛到对应真值.在实际应用中,工业对象可以利用慢时变线性模型表示,因此该递推算法能够确保工业对象模型实时更新以跟踪工况的变化.通过实例仿真可以看出,该递推算法能够保证慢时变线性模型参数有效更新,并较为准确估计输出变量.
 
 

关键词: 非线性, 慢时变, 线性模型, 递推算法, 有界收敛

Abstract:

The recursive algorithm for identifying the slow time-varying linear model was proposed, and its bounded convergence was analyzed. Base on the recursive algorithm, the parameters of slow time-varying linear model were proved to converge in bounded space which includes the collection of the true values of parameters. If working condition holds on, the parameters will converge to the corresponding true values with reasonable convergence factor. Generally, industrial plants can be described by slow time-varying linear model. Therefore, the model of indstrial plant can be updated in time with the recursive algorithm to track the characteristics changes effectively. As can be seen from the simulation example, the recursive algorithm can make sure that the parameters of slow time-varying linear model can be updated effectively and the output variables can be estimated accurately.
 

Key words: nonlinear, slow time-varying, linear model, recursive algorithm, bounded convergence

中图分类号: