Scholarlink

<p>In this article, a data-driven generalized iterative predictive control strategy is presented to address trajectory tracking challenges of unmanned aerial vehicle (UAV) operating in nonlinear, coupled, and disturbed environments. Within this framework, input–output (I/O) data are utilized to eliminate dependence on precise dynamic models and to enhance adaptability. A nominal dynamic linearization (DL) system is adopted to decouple pseudo-partial derivative (PPD) estimation from external disturbance. Disturbance effects are compensated through an iterative extended state observer (IESO). Computational efficiency comparable to the compact-form DL(CFDL) method is achieved. In addition, historical data, similar to those employed in partial-form and full-form DL methods, are incorporated to further improve control performance. Comparative simulations indicate significant improvements in tracking accuracy, convergence speed, and robustness over existing data-driven methods. The effectiveness of the proposed framework under dynamic conditions is demonstrated through experimental validation on the UAV system.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Data-Driven Generalized Iterative Predictive Control with IESO for UAV Trajectory Tracking

  • 期刊论文

  • 发表日期：

    2026年4月16日

查看全文
  • Liping Zhang,
  • Yujie Lin,
  • Yanjie Chen,
  • Ang Wang,
  • Zhiying Ren

摘要

In this article, a data-driven generalized iterative predictive control strategy is presented to address trajectory tracking challenges of unmanned aerial vehicle (UAV) operating in nonlinear, coupled, and disturbed environments. Within this framework, input–output (I/O) data are utilized to eliminate dependence on precise dynamic models and to enhance adaptability. A nominal dynamic linearization (DL) system is adopted to decouple pseudo-partial derivative (PPD) estimation from external disturbance. Disturbance effects are compensated through an iterative extended state observer (IESO). Computational efficiency comparable to the compact-form DL(CFDL) method is achieved. In addition, historical data, similar to those employed in partial-form and full-form DL methods, are incorporated to further improve control performance. Comparative simulations indicate significant improvements in tracking accuracy, convergence speed, and robustness over existing data-driven methods. The effectiveness of the proposed framework under dynamic conditions is demonstrated through experimental validation on the UAV system.