Scholarlink

<p>In this research, a nonsingular fast terminal sliding mode control (NFTSMC) using a fuzzy feature selection neural network (FFSNN) and a disturbance observer (DO) is developed to improve the property of tracking control of a DC–DC buck converter. The NFTSMC is derived to guarantee the convergence of the tracking error. Considering the adverse effects of system nonlinearities such as disturbances and uncertainties, the FFSNN and DO are adopted to approximate the nonlinear function and lumped disturbance, which are incorporated into the controller through real-time feedforward compensation. The proposed neural network combines the merits of the fuzzy neural network (FNN) and feature selection mechanism, and its output is adjusted to the optimal values by the adaptive estimator. Moreover, both network reconstructed error and external disturbance can be compensated through the DO estimate. Both simulation and experimental studies are implemented to illustrate the validity of the designed hybrid controller in different test conditions, showing higher tracking precision and faster transient responses than comparative method.</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

Nonsingular fast terminal sliding mode control of DC–DC buck converter using fuzzy neural network and disturbance observer

  • 期刊论文

  • 发表日期：

    2025年1月27日

查看全文
  • Zhiwei Liu,
  • Xiaoyu Gong,
  • Juntao Fei

摘要

In this research, a nonsingular fast terminal sliding mode control (NFTSMC) using a fuzzy feature selection neural network (FFSNN) and a disturbance observer (DO) is developed to improve the property of tracking control of a DC–DC buck converter. The NFTSMC is derived to guarantee the convergence of the tracking error. Considering the adverse effects of system nonlinearities such as disturbances and uncertainties, the FFSNN and DO are adopted to approximate the nonlinear function and lumped disturbance, which are incorporated into the controller through real-time feedforward compensation. The proposed neural network combines the merits of the fuzzy neural network (FNN) and feature selection mechanism, and its output is adjusted to the optimal values by the adaptive estimator. Moreover, both network reconstructed error and external disturbance can be compensated through the DO estimate. Both simulation and experimental studies are implemented to illustrate the validity of the designed hybrid controller in different test conditions, showing higher tracking precision and faster transient responses than comparative method.