A New Fault Feature Extraction Method for Rotating Machinery Based on Multiple Sensors.

Area of Science:

• Mechanical Engineering
• Signal Processing
• Vibration Analysis

Background:

• Rotating machinery generates complex vibration signals that are often aliased and noisy.
• Conventional methods struggle to separate these signals, hindering equipment condition monitoring and fault diagnosis.
• Existing blind source separation algorithms are ineffective in strong pulse noise environments.

Purpose of the Study:

• To develop an effective method for separating aliased vibration signals in strong pulse noise environments.
• To improve the accuracy of condition monitoring and fault diagnosis for rotating machinery.
• To provide a novel approach for vibration signal separation.

Main Methods:

• Vibration signals were de-noised using the median filter (MF) method.
• An improved joint approximate diagonalization of eigenmatrices (JADE) algorithm was employed for signal separation.
• The proposed method was validated through simulation and application to a hybrid rotor system.

Main Results:

• The median filter effectively reduced strong pulse noise in the vibration signals.
• The improved JADE algorithm successfully separated the aliased vibration sources from the de-noised signals.
• Simulation results confirmed the effectiveness of the combined MF and improved JADE approach.

Conclusions:

• The proposed method, combining median filtering and an improved JADE algorithm, is effective for separating vibration signals in strong pulse noise.
• This approach offers a significant advancement for condition monitoring and fault diagnosis in rotating machinery.
• A new, robust separation technique is provided for challenging vibration signal environments.