Scholarlink

<p>The Wiener systems demonstrate wide applicability in both industrial automation and biomedical signal processing fields. In this paper, a forgetting factor-based maximum likelihood stochastic gradient algorithm is proposed. Firstly, filtering techniques are incorporated to process measurement noise and disturbances during system identification, thereby enhancing model accuracy and robustness. Furthermore, the fixed-weight limitation inherent in conventional maximum likelihood algorithms is effectively addressed through a dynamic forgetting factor mechanism. This innovative approach enables adaptive adjustment of historical data weighting, achieving optimal balance between rapid parameter tracking capability and enhanced noise suppression performance. Finally, the simulations indicate that the proposed algorithm shows better estimations compared to the Recursive Least Squares (RLS) algorithm.</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

Wiener Systems Identification Using Forgetting Factor-Based Maximum Likelihood Stochastic Gradient Algorithm

  • 期刊论文

  • 发表日期：

    2025年10月25日

查看全文
  • Yangjing Shi,
  • Jing Liu,
  • Yangyang Chen,
  • Chengbing Zhao

摘要

The Wiener systems demonstrate wide applicability in both industrial automation and biomedical signal processing fields. In this paper, a forgetting factor-based maximum likelihood stochastic gradient algorithm is proposed. Firstly, filtering techniques are incorporated to process measurement noise and disturbances during system identification, thereby enhancing model accuracy and robustness. Furthermore, the fixed-weight limitation inherent in conventional maximum likelihood algorithms is effectively addressed through a dynamic forgetting factor mechanism. This innovative approach enables adaptive adjustment of historical data weighting, achieving optimal balance between rapid parameter tracking capability and enhanced noise suppression performance. Finally, the simulations indicate that the proposed algorithm shows better estimations compared to the Recursive Least Squares (RLS) algorithm.