全文下载:
202604019.pdf
文章编号:1672-6987(2026)04-0152-07;DOI:10. 16351/j. 1672-6987. 2026. 04. 019
李 倩 1,2 ,邢建民 1* (1. 青岛科技大学 数理学院,山东 青岛 266061;2. 临沂第十九中学,山东 临沂 276017)
摘 要:针对一类非线性非仿射离散时间系统,研究了随机批次变化的迭代学习控制问题,并 通过迭代方向上的动态线性化方法,提出了一种非线性数据驱动的范数最优 ILC 方法。非线 性学习控制律和非线性参数估计律均采用满足伯努利分布的随机变量来处理批次长度的变 化。此外,所提出的学习控制律是迭代时变的,在同一次迭代中利用了更多以前时刻的控制 信息,从而使控制性能更加有效。该方法是数据驱动的,不使用任何进程模型信息,只使用 I/ O 数据。在伯努利分布假设下,严格地证明了在概率意义上的收敛性分析。通过数值模拟对 理论结果进行了验证。
关键词:数据驱动控制;迭代时变学习增益;可变批次长度;非线性非仿射系统
中图分类号:TP 273 文献标志码:A
引用格式:李倩,邢建民 . 具有随机批次长度的数据驱动范数最优迭代学习控制[J]. 青岛科 技大学学报(自然科学版),2026,47(4):152-158.
LI Qian, XING Jianmin. Data-driven norm-optimal ILC with varying trial lengths[J]. Journal of Qingdao University of Science and Technology(Natural Science Edition),2026,47(4): 152-158.
Data-Driven Norm-Optimal ILC with Varying Trial Lengths
LI Qian1,2 , XING Jianmin1 (1. College of Mathematics and Physics, Qingdao University of Science and Technology, Qingdao 266061, China; 2. Linyi No. 19 Middle School, Linyi 276017, China)
Abstract:This work deals with randomly varying trial lengths of a nonlinear and non-affine discrete-time system and proposes a nonlinear data-driven norm-optimal ILC method via a dynamic linearization method in the iteration direction. Both the nonlinear learning control law and the nonlinear parameter estimation law are performed by using a stochastic variable satisfy⁃ ing Bernoulli distribution to handle the varying trial lengths. Further, the learning control law proposed is iteration-time-varying and makes use of more control information at previous time instants within the same iteration such that the control performance can be more effectively. The proposed method is data-driven and does not use any process model information but the I/ O data. Under the Bernoulli distribution assumption, the convergence analysis is demonstrated rigorously in the probability sense. Some numerical simulations have been provided to verify the theoretical results.
Key words:data-driven control; norm optimal ILC; iteration-time-varying learning gain; varying trial length; nonlinear non-affine systems
收稿日期:2025-08-05
基金项目:青岛科技大学科研启动项目(12030430010187).
作者简介:李 倩(1996—),女,硕士研究生 . * 通信联系人 .