Rotating Machinery Diagnosing in Non-Stationary Conditions with Empirical Mode Decomposition-Based Wavelet Leaders...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Non-invasive Diagnostics

Background:

• Rotating machine diagnostics rely on analyzing dynamic sensor signals (vibration, acceleration, etc.).
• Non-stationary conditions pose challenges for traditional fault detection methods.
• Damage progression often alters the signal's dynamic characteristics.

Purpose of the Study:

• To present a novel method for diagnosing rotating machine damage.
• To apply empirical mode decomposition and wavelet leader multifractal analysis to fault detection.
• To assess the method's effectiveness under non-stationary operating conditions.

Main Methods:

• Combining empirical mode decomposition (EMD) with wavelet leader multifractal formalism.
• Analyzing dynamic signals (vibration, acceleration) from sensors.
• Classifying features of the multifractal spectrum to diagnose faults.

Main Results:

• Increased multifractality indicates developing damage in rotating machines.
• The multifractal spectrum's shape changes diagnostically with fault progression.
• The method effectively identifies faults causing impulse responses in signals.

Conclusions:

• The proposed EMD and wavelet multifractal method offers a robust approach to diagnosing rotating machine faults.
• This technique is particularly effective for detecting impulse-generating faults under non-stationary conditions.
• Validation on laboratory and real-world vibration data confirms the method's practical applicability.