Scholarlink

<p>The analysis of variable-speed signals is challenged by issues such as harmonic interference and amplitude fluctuations, yet the most fundamental problem stems from amplitude-frequency modulation (AM-FM) coupling induced by speed variations. Conventional Fourier-transform-based methods are unable to accurately characterize such coupling, which in turn limits the performance of instantaneous frequency (IF) estimation techniques that depend on high-precision harmonic information. To overcome these limitations, we propose a squared amplitude spectrum (SAS) to strengthen time-frequency representations, enabling robust IF estimation under strong modulation and adjacent interference through enhanced emphasis on dominant frequency components and concurrent noise suppression. A more significant contribution of this work is a weighted maximum likelihood estimation (WMLE) framework, designed with localized adaptive weights derived from the normalized power spectral density to precisely track time-varying signal structures. By iteratively optimizing a closed-form likelihood function, the framework continuously refines initial IF estimates. Moreover, an embedded signal-to-noise ratio (SNR)-weighted mechanism provides adaptable control over the trade-off between noise suppression and resolution enhancement, ensuring consistent performance across varied operational conditions. The proposed SAS method achieves an average root mean square error (RMSE) value 83 % lower than that of the short time Fourier transform (STFT) method under various SNR in simulated signal. And in both simulated signal with different noise levels and experimental cases involving the Ottawa bearing and RV reducer crankshaft bearing under diverse operating conditions, the proposed method consistently yields the lowest average IF estimation errors. in RMSE over the best performing baseline 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

SAS-WMLE-optimized instantaneous frequency estimation for variable-speed signals in RV reducers

  • 期刊论文

  • 发表日期：

    2026年7月11日

查看全文
  • Guangqi Qiu,
  • Yan Chen,
  • Wenjuan Xie,
  • Guangjun Jiang,
  • Peng Huang,
  • Yingkui Gu

摘要

The analysis of variable-speed signals is challenged by issues such as harmonic interference and amplitude fluctuations, yet the most fundamental problem stems from amplitude-frequency modulation (AM-FM) coupling induced by speed variations. Conventional Fourier-transform-based methods are unable to accurately characterize such coupling, which in turn limits the performance of instantaneous frequency (IF) estimation techniques that depend on high-precision harmonic information. To overcome these limitations, we propose a squared amplitude spectrum (SAS) to strengthen time-frequency representations, enabling robust IF estimation under strong modulation and adjacent interference through enhanced emphasis on dominant frequency components and concurrent noise suppression. A more significant contribution of this work is a weighted maximum likelihood estimation (WMLE) framework, designed with localized adaptive weights derived from the normalized power spectral density to precisely track time-varying signal structures. By iteratively optimizing a closed-form likelihood function, the framework continuously refines initial IF estimates. Moreover, an embedded signal-to-noise ratio (SNR)-weighted mechanism provides adaptable control over the trade-off between noise suppression and resolution enhancement, ensuring consistent performance across varied operational conditions. The proposed SAS method achieves an average root mean square error (RMSE) value 83 % lower than that of the short time Fourier transform (STFT) method under various SNR in simulated signal. And in both simulated signal with different noise levels and experimental cases involving the Ottawa bearing and RV reducer crankshaft bearing under diverse operating conditions, the proposed method consistently yields the lowest average IF estimation errors. in RMSE over the best performing baseline method.