Scholarlink

<p>Underwater acoustic signals are severely corrupted by non-Gaussian impulsive noise with heavy-tailed distribution, which significantly degrades conventional adaptive filtering methods. To address this issue, a novel fractional lower-order adaptive filtering algorithm based on arctangent nonlinear transformation (ARLMP) is proposed for underwater acoustic signal denoising. The core innovation of the proposed ARLMP lies in integrating the minimum dispersion criterion with the saturation property of the arctangent function. This dynamically constrains the error signal, suppresses impulsive outliers on weight updates, and preserves critical signal structures. The performance of ARLMP is validated on real underwater acoustic signals from the ShipsEar dataset contaminated with mixed Wenz spectrum noise and alpha-stable impulsive noise. Comprehensive comparisons with the Variational Mode Decomposition (VMD) and Normalized Least Mean <i>p</i>-Power (NLMP) algorithms are conducted in the time, frequency and time-frequency domains. Experimental results demonstrate that ARLMP consistently achieves the highest output signal-to-noise ratio(SNR) under all test conditions, with robust denoising performance, superior signal fidelity, and high computational efficiency. This study provides an effective solution for impulsive noise suppression in complex underwater acoustic environments.</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

Arctangent-Based Robust Adaptive Filtering for Underwater Acoustic Signal Denoising in Impulsive Environments

  • 期刊论文

  • 发表日期：

    2026年3月10日

查看全文
  • Jiang Yulin,
  • Gao Song,
  • Ding Wenbin

摘要

Underwater acoustic signals are severely corrupted by non-Gaussian impulsive noise with heavy-tailed distribution, which significantly degrades conventional adaptive filtering methods. To address this issue, a novel fractional lower-order adaptive filtering algorithm based on arctangent nonlinear transformation (ARLMP) is proposed for underwater acoustic signal denoising. The core innovation of the proposed ARLMP lies in integrating the minimum dispersion criterion with the saturation property of the arctangent function. This dynamically constrains the error signal, suppresses impulsive outliers on weight updates, and preserves critical signal structures. The performance of ARLMP is validated on real underwater acoustic signals from the ShipsEar dataset contaminated with mixed Wenz spectrum noise and alpha-stable impulsive noise. Comprehensive comparisons with the Variational Mode Decomposition (VMD) and Normalized Least Mean p-Power (NLMP) algorithms are conducted in the time, frequency and time-frequency domains. Experimental results demonstrate that ARLMP consistently achieves the highest output signal-to-noise ratio(SNR) under all test conditions, with robust denoising performance, superior signal fidelity, and high computational efficiency. This study provides an effective solution for impulsive noise suppression in complex underwater acoustic environments.