Intelligent Diagnosis Method for Rotating Machinery Using Dictionary Learning and Singular Value Decomposition.

Area of Science:

• Engineering
• Mechanical Engineering
• Signal Processing

Background:

• Rotating machinery is crucial in industry but prone to failures under demanding conditions.
• Condition monitoring and fault diagnosis (CMFD) are essential for ensuring operational reliability and safety.
• Existing methods may lack adaptability and capacity for complex fault signatures.

Purpose of the Study:

• To propose an intelligent fault diagnosis method for rotating machinery.
• To leverage dictionary learning (DL) and singular value decomposition (SVD) for adaptive feature extraction.
• To enhance the accuracy and efficiency of fault pattern classification.

Main Methods:

• Dictionary learning is employed for adaptive feature extraction from raw signals without prior knowledge.
• Singular value decomposition (SVD) is used to derive feature vectors from the learned dictionary matrix.
• Principal component analysis (PCA) reduces feature vector dimensionality, followed by K-nearest neighbor (KNN) for classification.

Main Results:

• The proposed method successfully extracts underlying signal structures using dictionary learning.
• Feature vectors derived from SVD and reduced by PCA enable effective fault pattern identification.
• Experimental case studies confirm the method's effectiveness in diagnosing rotating machinery faults.

Conclusions:

• The dictionary learning-based approach demonstrates superior capacity and adaptability for feature extraction compared to mode decomposition methods.
• The integrated DL-SVD-PCA-KNN framework provides an effective solution for intelligent fault diagnosis.
• This method enhances the reliability and security of rotating machinery operations.