Scholarlink

<p>To address the challenges posed by traditional contact-based measurements and the susceptibility of conventional ridge extraction algorithms to local optima under low signal-to-noise ratios in variable-speed bearing fault diagnosis, a non-contact diagnostic method integrating visual perception with an entropy-enhanced strategy is proposed in this study. Initially, vibration signals are extracted from high-speed video sequences through the Farneback dense optical flow method, achieving non-contact signal acquisition. Subsequently, an enhanced Crazy Climber algorithm, optimized by the maximum entropy principle (MEPCC), is introduced. This advancement incorporates entropy-energy co-optimization and an adaptive phase-shifting mechanism, leading to improved robustness and continuity of ridge extraction. Numerical simulations and experimental case studies show that, under the tested conditions, the proposed method can extract relatively smooth and continuous time–frequency ridges under strong noise interference, thereby supporting order-based fault identification in variable-speed bearings. A feasible non-contact solution for the intelligent diagnostics of rotating machinery is thus provided.</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

Non-contact fault diagnosis of variable speed bearings based on optical flow method and entropy enhanced ridge extraction

  • 期刊论文

  • 发表日期：

    2026年8月4日

查看全文
  • Xu Zhang,
  • Jun Zhou

摘要

To address the challenges posed by traditional contact-based measurements and the susceptibility of conventional ridge extraction algorithms to local optima under low signal-to-noise ratios in variable-speed bearing fault diagnosis, a non-contact diagnostic method integrating visual perception with an entropy-enhanced strategy is proposed in this study. Initially, vibration signals are extracted from high-speed video sequences through the Farneback dense optical flow method, achieving non-contact signal acquisition. Subsequently, an enhanced Crazy Climber algorithm, optimized by the maximum entropy principle (MEPCC), is introduced. This advancement incorporates entropy-energy co-optimization and an adaptive phase-shifting mechanism, leading to improved robustness and continuity of ridge extraction. Numerical simulations and experimental case studies show that, under the tested conditions, the proposed method can extract relatively smooth and continuous time–frequency ridges under strong noise interference, thereby supporting order-based fault identification in variable-speed bearings. A feasible non-contact solution for the intelligent diagnostics of rotating machinery is thus provided.