Scholarlink

<p>This paper proposes an adaptive event-triggered online learning control scheme for uncertain strict-feedback systems. Existing event-triggered online learning methods typically rely on prior knowledge of the Lipschitz constants or upper bounds of the Reproducing Kernel Hilbert Space norms of system dynamics, which are often unavailable in practice. To overcome this issue, an adaptive law is developed to estimate the Lipschitz constants in real time. Furthermore, an adaptive event-triggered mechanism is designed, ensuring that the Gaussian process model updates training data only when triggering conditions are violated, thereby enhancing data efficiency and reducing computational complexity. Through Lyapunov stability theory, we prove that the proposed online learning control algorithm ensures the boundedness of all signals in the closed-loop system. The effectiveness and practical applicability of the proposed method are demonstrated through numerical simulations and real-time experiments on a Franka Emika Panda robotic arm.</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

Adaptive event-triggered online learning control of uncertain strict-feedback systems using Gaussian processes: theory and experiments

  • 期刊论文

  • 发表日期：

    2026年4月8日

查看全文
  • Yu-Fa Liu,
  • Lin-Feng Huang,
  • Gui-Ming Lian,
  • Jie Tao,
  • Min Lin,
  • Yong-Hua Liu,
  • Chun-Yi Su,
  • Renquan Lu

摘要

This paper proposes an adaptive event-triggered online learning control scheme for uncertain strict-feedback systems. Existing event-triggered online learning methods typically rely on prior knowledge of the Lipschitz constants or upper bounds of the Reproducing Kernel Hilbert Space norms of system dynamics, which are often unavailable in practice. To overcome this issue, an adaptive law is developed to estimate the Lipschitz constants in real time. Furthermore, an adaptive event-triggered mechanism is designed, ensuring that the Gaussian process model updates training data only when triggering conditions are violated, thereby enhancing data efficiency and reducing computational complexity. Through Lyapunov stability theory, we prove that the proposed online learning control algorithm ensures the boundedness of all signals in the closed-loop system. The effectiveness and practical applicability of the proposed method are demonstrated through numerical simulations and real-time experiments on a Franka Emika Panda robotic arm.