Scholarlink

Purpose <p>The synchrosqueezing transform (SST), fractional synchrosqueezing transformation (FRSST) and multi-synchrosqueezing transform (MSST) use window functions with constant window sizes to extract instantaneous fault frequency (IFF). Under conditions of large speed fluctuations, the time–frequency aggregation and estimation accuracy of vibration signals are poor and low. To solve the above problems, an adaptive window length fractional synchrosqueezing transform (AWLFRSST) algorithm is proposed.</p> Methods <p>Firstly, considering the local variation characteristics of vibration signals under speed fluctuation conditions, a time-varying Gaussian window function is introduced into FRSST. The theoretical derivation of AWLFRSST and the estimation steps of instantaneous fault characteristic frequency are given. Secondly, using local Renyi entropy as a performance index, the optimal time-varying window length function is estimated to locally optimize the time–frequency aggregation and instantaneous frequency estimation accuracy. Finally, the AWLFRSST is used to simulated signals and actual wheelset bearing vibration signals.</p> Results <p>For variable-speed rolling bearing vibration signals with inner ring faults, AWLFRSST reduced the Renyi entropy of the time–frequency distribution by 16.97% and 11.99% compared to MSST and FRSST, while decreasing IFF extraction errors by 56.48% and 34.90%. For signals containing outer ring faults, it achieved reductions of 16.89% and 11.11% in Renyi entropy, and 59.02% and 39.29% in IFF extraction errors relative to MSST and FRSST respectively.</p> Conclusions <p>The AWLFRSST method exhibits superior time–frequency aggregation, higher IFF extraction accuracy, and stronger noise robustness when processing nonlinear variable speed vibration signals.</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

The Extraction of Instantaneous Fault Frequency of Wheelset Bearings Based on Adaptive Window Length Fractional Synchrosqueezing Transform

  • 期刊论文

  • 发表日期：

    2025年9月18日

查看全文
  • Xin Li,
  • Zengqiang Ma,
  • Zeliang Ma,
  • Xiaoheng Zhang

摘要

Purpose

The synchrosqueezing transform (SST), fractional synchrosqueezing transformation (FRSST) and multi-synchrosqueezing transform (MSST) use window functions with constant window sizes to extract instantaneous fault frequency (IFF). Under conditions of large speed fluctuations, the time–frequency aggregation and estimation accuracy of vibration signals are poor and low. To solve the above problems, an adaptive window length fractional synchrosqueezing transform (AWLFRSST) algorithm is proposed.

Methods

Firstly, considering the local variation characteristics of vibration signals under speed fluctuation conditions, a time-varying Gaussian window function is introduced into FRSST. The theoretical derivation of AWLFRSST and the estimation steps of instantaneous fault characteristic frequency are given. Secondly, using local Renyi entropy as a performance index, the optimal time-varying window length function is estimated to locally optimize the time–frequency aggregation and instantaneous frequency estimation accuracy. Finally, the AWLFRSST is used to simulated signals and actual wheelset bearing vibration signals.

Results

For variable-speed rolling bearing vibration signals with inner ring faults, AWLFRSST reduced the Renyi entropy of the time–frequency distribution by 16.97% and 11.99% compared to MSST and FRSST, while decreasing IFF extraction errors by 56.48% and 34.90%. For signals containing outer ring faults, it achieved reductions of 16.89% and 11.11% in Renyi entropy, and 59.02% and 39.29% in IFF extraction errors relative to MSST and FRSST respectively.

Conclusions

The AWLFRSST method exhibits superior time–frequency aggregation, higher IFF extraction accuracy, and stronger noise robustness when processing nonlinear variable speed vibration signals.