Scholarlink

Abstract <p>Acoustic source localization plays a vital role in fault diagnosis, and time difference of arrival (TDoA) is an important acoustic source localization method. In engineering practice, non-stationary working conditions and strong noise background bring challenges to TDoA and fault identification. In this paper, a novel acoustic source localization method called ultra-narrow band modal decomposition (UNBMD) based TDoA is proposed. First, frequency-domain windows corresponding to characteristic modes is constructed to enhance the decomposition accuracy of signal under strong noise. Second, correlation analysis is utilized to obtain accurate time delays for each channel. In the simulation studies, signal-to-noise ratio (SNR) and waveform similarity are adopted as evaluation metrics. Comparative analysis with variational mode decomposition and band-pass filtering methods demonstrates that the proposed approach achieves a waveform similarity of 99.86% and an SNR of 25.3427 dB, thereby exhibiting superior performance. In engineering verification, an experimental study was conducted on actual tobacco machinery. The results indicate that the proposed UNBMD-TDoA method can realize the acoustic source localization under strong noise conditions, offering a reliable technical foundation for the fault diagnosis and maintenance of mechanical systems.</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

Acoustic Source Localization Based on Ultra-Narrow Band Modal Decomposition under Nonstationary Operating Conditions

  • 期刊论文

  • 发表日期：

    2026年2月1日

查看全文
  • Li-ao Chen,
  • Qingqi Li,
  • Chuanwei Li,
  • Youchao Lu,
  • Guoyun Huang,
  • Hanzhang Huang,
  • Wenkui Zhu,
  • Xiaofeng Du

摘要

Abstract

Acoustic source localization plays a vital role in fault diagnosis, and time difference of arrival (TDoA) is an important acoustic source localization method. In engineering practice, non-stationary working conditions and strong noise background bring challenges to TDoA and fault identification. In this paper, a novel acoustic source localization method called ultra-narrow band modal decomposition (UNBMD) based TDoA is proposed. First, frequency-domain windows corresponding to characteristic modes is constructed to enhance the decomposition accuracy of signal under strong noise. Second, correlation analysis is utilized to obtain accurate time delays for each channel. In the simulation studies, signal-to-noise ratio (SNR) and waveform similarity are adopted as evaluation metrics. Comparative analysis with variational mode decomposition and band-pass filtering methods demonstrates that the proposed approach achieves a waveform similarity of 99.86% and an SNR of 25.3427 dB, thereby exhibiting superior performance. In engineering verification, an experimental study was conducted on actual tobacco machinery. The results indicate that the proposed UNBMD-TDoA method can realize the acoustic source localization under strong noise conditions, offering a reliable technical foundation for the fault diagnosis and maintenance of mechanical systems.