Robust post-processing time frequency technology and its application to mechanical fault diagnosis.

Area of Science:

• Signal Processing
• Mechanical Engineering
• Fault Diagnosis

Background:

• Traditional synchrosqueezing transform (SST) and synchroextracting transform (SET) methods enhance time-frequency resolution (TFR) for fault diagnosis.
• Normal and fault signals, including background noise, can be modeled as infinite variance processes (1 < α ≤ 2), challenging conventional TFR methods.
• The efficacy of standard SST and SET is significantly diminished in infinite variance process environments.

Purpose of the Study:

• To develop robust post-processing methods for improving TFR resolution in fault diagnosis under infinite variance process conditions.
• To mathematically derive and validate novel algorithms designed to overcome the limitations of traditional methods in complex signal environments.

Main Methods:

• Proposed robust post-processing techniques: FSET, FSSET, FSOSET, and FMSST.
• Utilized infinite variance process statistical models and the FLOS technique for algorithm development.
• Completed mathematical derivations for the proposed methods.

Main Results:

• The proposed methods demonstrate superior performance compared to conventional SST and SET techniques.
• Applied to diagnose bearing outer race signals corrupted by infinite variance processes, the new methods show significant performance advantages.
• Comparative analysis confirms the enhanced TFR resolution and diagnostic accuracy of the novel algorithms.

Conclusions:

• The developed FSET, FSSET, FSOSET, and FMSST methods offer robust solutions for fault diagnosis in signals characterized by infinite variance processes.
• These advanced algorithms provide improved TFR resolution and diagnostic performance, particularly in noisy and complex signal conditions.
• The study summarizes the characteristics, limitations, and application scenarios of the improved algorithms for future research and industrial implementation.