Scholarlink

<p>The Volterra series filtering model is most prevalent in nonlinear systems. However, when non-Gaussian noise exists in the environment, the parameter identification ability of the Volterra series algorithm is greatly affected, and the model has a high computational cost. Therefore, this paper proposes a data-selective Volterra filtering algorithm based on variable correntropy with impulse detection capability. The filter coefficients stop updating when the algorithm detects impulse noise or identifies input data that does not bring enough novelty to the system. Therefore, the proposed algorithm can effectively improve the resistance to impulse noise, improve the parameter identification accuracy of nonlinear systems, and reduce the computational cost to some extent. Simulation results show that the proposed algorithm performs robustly in the nonlinear system identification problem and is not affected by changes in environmental noise. And the update rate of the filter coefficients is only 41%, which greatly reduces the computational cost of the Volterra series filter 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

Robust Data-Selective Nonlinear System Identification Based on Volterra Model

  • 期刊论文

  • 发表日期：

    2025年5月7日

查看全文
  • Lingjie Sheng,
  • Yaowei Guo,
  • Junhui Qian,
  • Guobing Qian

摘要

The Volterra series filtering model is most prevalent in nonlinear systems. However, when non-Gaussian noise exists in the environment, the parameter identification ability of the Volterra series algorithm is greatly affected, and the model has a high computational cost. Therefore, this paper proposes a data-selective Volterra filtering algorithm based on variable correntropy with impulse detection capability. The filter coefficients stop updating when the algorithm detects impulse noise or identifies input data that does not bring enough novelty to the system. Therefore, the proposed algorithm can effectively improve the resistance to impulse noise, improve the parameter identification accuracy of nonlinear systems, and reduce the computational cost to some extent. Simulation results show that the proposed algorithm performs robustly in the nonlinear system identification problem and is not affected by changes in environmental noise. And the update rate of the filter coefficients is only 41%, which greatly reduces the computational cost of the Volterra series filter algorithm.