A novel fault classification feature extraction method for rolling bearing based on multi-sensor fusion technology...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Machine Learning

Background:

• Accurate bearing fault diagnosis is crucial in multisensor monitoring environments.
• Existing methods often struggle to extract effective fault classification features.
• Need for enhanced feature extraction for improved diagnostic accuracy.

Purpose of the Study:

• To propose a novel bearing fault classification feature extraction method.
• To enhance the accuracy and speed of rolling bearing fault diagnosis.
• To leverage multisensor fusion technology and an enhanced binary one-dimensional ternary pattern (EB-1D-TP) algorithm.

Main Methods:

• Developed an optimal equalization weighting algorithm for high-precision signal fusion.
• Introduced an enhanced binary encoding method (similar to balanced ternary encoding) to increase feature differentiation.
• Utilized a support vector machine (SVM) for fault classification using the fused and encoded features.

Main Results:

• The proposed algorithm significantly improved the accuracy of rolling bearing fault classification.
• The method demonstrated a notable increase in the speed of fault classification.
• Combining fusion-encoding features with other intelligent classifiers further enhanced diagnostic outcomes.

Conclusions:

• The multisensor fusion and EB-1D-TP algorithm provide a powerful tool for bearing fault diagnosis.
• The enhanced feature extraction method leads to more accurate and efficient fault identification.
• This approach offers a promising direction for improving condition monitoring systems.