Scholarlink

<p>The electric drive system (EDS) represents a sophisticated nonlinear power mechanism, necessitating a high level of reliability and stability. Condition monitoring often relies on multi-scale coarse-grained analysis techniques. However, challenges such as time domain weaknesses and frequency aliasing can hinder accurate weak fault detection. To address this, a novel method known as improved fine multi-scale reverse weighted permutation entropy (IFMRWPE) has been introduced to extract nonlinear fault features. The IFMRWPE method initially filters vibration signals across different scales to reduce frequency aliasing effects. Subsequently, it eliminates large-scale features, capturing fine entropy values that delineate subtle vibration impacts. An experimental study on the EDS, specifically focusing on bearing faults, was carried out in a controlled laboratory environment to acquire diverse signal qualities. The study delved into the identification attributes of three common bearing faults within EDS, employing machine learning for fault prediction assessment. The outcomes revealed that IFMRWPE-PSO-BPNN demonstrates a robust weak fault identification capability, boasting approximately 93% diagnostic accuracy. This underscores its efficacy and broad applicability in enhancing fault prediction within electric drive 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

A bearing fault diagnosis method for electric drive systems based on improved fine multi-scale reverse weighted permutation entropy

  • 期刊论文

  • 发表日期：

    2026年3月5日

查看全文
  • Cong Yue,
  • Jiabin Dong,
  • Bei Wang,
  • Kai Wang,
  • Ping Cheng

摘要

The electric drive system (EDS) represents a sophisticated nonlinear power mechanism, necessitating a high level of reliability and stability. Condition monitoring often relies on multi-scale coarse-grained analysis techniques. However, challenges such as time domain weaknesses and frequency aliasing can hinder accurate weak fault detection. To address this, a novel method known as improved fine multi-scale reverse weighted permutation entropy (IFMRWPE) has been introduced to extract nonlinear fault features. The IFMRWPE method initially filters vibration signals across different scales to reduce frequency aliasing effects. Subsequently, it eliminates large-scale features, capturing fine entropy values that delineate subtle vibration impacts. An experimental study on the EDS, specifically focusing on bearing faults, was carried out in a controlled laboratory environment to acquire diverse signal qualities. The study delved into the identification attributes of three common bearing faults within EDS, employing machine learning for fault prediction assessment. The outcomes revealed that IFMRWPE-PSO-BPNN demonstrates a robust weak fault identification capability, boasting approximately 93% diagnostic accuracy. This underscores its efficacy and broad applicability in enhancing fault prediction within electric drive systems.