Scholarlink

<p>This paper studies the issues of sampled-data (SD) tracking control and performance optimization for nonlinear systems subject to input saturation, via Takagi-Sugeno (T-S) fuzzy method and reinforcement learning. The aim is to construct an SD fuzzy controller to drive the states of T-S fuzzy systems subject to input saturation to track the reference model’s states. This control design adopts the imperfect premise matching (IPM) concept to allow membership functions (MFs) of the SD fuzzy controller to differ from MFs of the T-S fuzzy model, which effectively resolves the mismatch problem of MFs induced by the simultaneous consideration of the SD control and optimization of MFs. The sector nonlinearity technique is employed for handling the nonconvex issue arising from input saturation. To achieve less conservative results, this paper introduces the predetermined information of MFs into the analysis through the MFs-dependent (MFD) analysis approach. Afterwards, an optimization strategy based on the deep deterministic policy gradient (DDPG) algorithm is proposed to optimize MFs of the SD fuzzy controller. Finally, to verify the effectiveness of the proposed method, simulation results are presented.</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

Sampled-data tracking control for fuzzy systems subject to input saturation with optimization of membership functions

  • 期刊论文

  • 发表日期：

    2025年12月3日

查看全文
  • Ming Chen,
  • Xicheng Yang,
  • Hak-Keung Lam,
  • Zhe Sun,
  • Ying Shen,
  • Xiangming Ye

摘要

This paper studies the issues of sampled-data (SD) tracking control and performance optimization for nonlinear systems subject to input saturation, via Takagi-Sugeno (T-S) fuzzy method and reinforcement learning. The aim is to construct an SD fuzzy controller to drive the states of T-S fuzzy systems subject to input saturation to track the reference model’s states. This control design adopts the imperfect premise matching (IPM) concept to allow membership functions (MFs) of the SD fuzzy controller to differ from MFs of the T-S fuzzy model, which effectively resolves the mismatch problem of MFs induced by the simultaneous consideration of the SD control and optimization of MFs. The sector nonlinearity technique is employed for handling the nonconvex issue arising from input saturation. To achieve less conservative results, this paper introduces the predetermined information of MFs into the analysis through the MFs-dependent (MFD) analysis approach. Afterwards, an optimization strategy based on the deep deterministic policy gradient (DDPG) algorithm is proposed to optimize MFs of the SD fuzzy controller. Finally, to verify the effectiveness of the proposed method, simulation results are presented.