Scholarlink

Purpose <p>To evaluate the influence of different types of data window selection, applied to time series recording resonance regimes.</p> Method <p>Spectra were calculated by windowing vibration signals. Four window types are used: <i>Rectangular</i>, <i>Hannin</i>g, <i>Hamming</i> and <i>Blackman</i>. For all of them, two parameters: window type/shape and length <i>L</i>, which influence its spectrum and consequently the spectral analysis, were assessed.</p> Results <p>Regarding the windows shape, no important variations by using Hanning, Blackman or Hamming were noted. For signal with the presence of resonance (chirp-like signal), the Rectangular window is not recommended because of the larger sidelobes and do not allow the analysis of the isolated resonance frequency. For smaller window length <i>L</i>, the measured resonance is in smaller frequencies, due to the spectral leakage effect. On the other side, for larger values of <i>L</i>, more frequencies are analyzed, what can prevent the definition the frequency of resonance. For the sampling rate used to collect the source signal, values of <i>L</i> around 1000 is the best choice.</p> Conclusions <p>The <i>Hanning</i> and <i>Blackman</i> windows present very similar results and, despite the wider mainlobe when compared to the <i>Rectangular</i> window, the smaller levels and faster decay sidelobes contribute to more accurate determination of the resonance frequencies. Due to the chirp-like excitation signal and resonance regime, the window length must be determined according to the sampling rate and rate of change of the chirp signal. Small window lengths have the problem of wide mainlobes, and for larger window lengths, more frequencies are windowed, and a trade-off must be determined for each system.</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

Study of Windowing in Spectral Analysis of Resonance Regimes

  • 期刊论文

  • 发表日期：

    2025年9月26日

查看全文
  • Alexandre de Macêdo Wahrhaftig,
  • Ricardo Tokio Higuti,
  • Iryna Bondarenko,
  • Larysa Neduzha

摘要

Purpose

To evaluate the influence of different types of data window selection, applied to time series recording resonance regimes.

Method

Spectra were calculated by windowing vibration signals. Four window types are used: Rectangular, Hanning, Hamming and Blackman. For all of them, two parameters: window type/shape and length L, which influence its spectrum and consequently the spectral analysis, were assessed.

Results

Regarding the windows shape, no important variations by using Hanning, Blackman or Hamming were noted. For signal with the presence of resonance (chirp-like signal), the Rectangular window is not recommended because of the larger sidelobes and do not allow the analysis of the isolated resonance frequency. For smaller window length L, the measured resonance is in smaller frequencies, due to the spectral leakage effect. On the other side, for larger values of L, more frequencies are analyzed, what can prevent the definition the frequency of resonance. For the sampling rate used to collect the source signal, values of L around 1000 is the best choice.

Conclusions

The Hanning and Blackman windows present very similar results and, despite the wider mainlobe when compared to the Rectangular window, the smaller levels and faster decay sidelobes contribute to more accurate determination of the resonance frequencies. Due to the chirp-like excitation signal and resonance regime, the window length must be determined according to the sampling rate and rate of change of the chirp signal. Small window lengths have the problem of wide mainlobes, and for larger window lengths, more frequencies are windowed, and a trade-off must be determined for each system.