Scholarlink

Purpose <p>The flexible thin-walled elliptical bearing (FTEB), a key component of harmonic reducers, produces elliptical periodic impacts that mask fault-induced impulses, rendering conventional diagnostic methods ineffective. This study aims to develop a novel signal processing approach to overcome this limitation and enable accurate weak fault detection in FTEBs.</p> Methods <p>A novel auxiliary-signal-guided principal component analysis (PCA) approach is proposed. Unlike traditional PCA and signal decomposition techniques, the method exploits two theoretical properties: (i) each frequency component generates a pair of adjacent nonzero eigenvalues, and (ii) eigenvalue ranking follows amplitude dominance. By injecting a sinusoidal component at the target frequency, the eigenvalue distribution is reshaped, enabling precise localization and separation of weak fault features.</p> Results <p>Simulation results demonstrate the superior capability of the proposed method in isolating single-frequency components. Experimental analysis of FTEB signals confirms accurate extraction of fault characteristic frequencies and their harmonics, free from interference caused by elliptical impacts<b>.</b></p> Conclusion <p>The proposed method provides a robust diagnostic tool with enhanced frequency resolution and feature separation performance, offering new insight into weak fault detection for FTEBs.</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 Novel Principal Component Analysis Method Guided by Auxiliary Target Signals and its Application to Fault Diagnosis of Flexible Thin-walled Elliptical Bearings

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Mingjun Guo,
  • Jiajia Pan,
  • Xuezhi Zhao

摘要

Purpose

The flexible thin-walled elliptical bearing (FTEB), a key component of harmonic reducers, produces elliptical periodic impacts that mask fault-induced impulses, rendering conventional diagnostic methods ineffective. This study aims to develop a novel signal processing approach to overcome this limitation and enable accurate weak fault detection in FTEBs.

Methods

A novel auxiliary-signal-guided principal component analysis (PCA) approach is proposed. Unlike traditional PCA and signal decomposition techniques, the method exploits two theoretical properties: (i) each frequency component generates a pair of adjacent nonzero eigenvalues, and (ii) eigenvalue ranking follows amplitude dominance. By injecting a sinusoidal component at the target frequency, the eigenvalue distribution is reshaped, enabling precise localization and separation of weak fault features.

Results

Simulation results demonstrate the superior capability of the proposed method in isolating single-frequency components. Experimental analysis of FTEB signals confirms accurate extraction of fault characteristic frequencies and their harmonics, free from interference caused by elliptical impacts.

Conclusion

The proposed method provides a robust diagnostic tool with enhanced frequency resolution and feature separation performance, offering new insight into weak fault detection for FTEBs.