Fault Diagnosis for Rolling Bearings under Variable Conditions Based on Visual Cognition

Area of Science:

• Mechanical Engineering
• Cognitive Computing
• Signal Processing

Background:

• Rolling bearing fault diagnosis is critical for industrial machinery maintenance.
• Existing methods often struggle with variable operating conditions.
• A need exists for robust fault detection systems that mimic human visual perception.

Purpose of the Study:

• To develop an effective fault diagnosis method for rolling bearings under variable conditions.
• To leverage principles of human visual cognition for improved feature extraction and classification.
• To provide a reliable approach for machinery health monitoring in dynamic environments.

Main Methods:

• Vibration signals are converted into recurrence plots (RPs) - 2D images.
• Speeded Up Robust Features (SURF) are used for translation, rotation, and scale-invariant feature extraction.
• Isometric mapping reduces feature dimensionality while preserving intrinsic data structure.
• Support Vector Machine (SVM) classifies bearing conditions.

Main Results:

• The proposed visual cognition-based method demonstrates high effectiveness in diagnosing rolling bearing faults under variable conditions.
• Experimental validation using Case Western Reserve University Bearing Data Center data confirms the method's efficacy.
• The approach achieves robust fault identification despite changes in operating parameters.

Conclusions:

• The developed fault diagnosis method offers a promising cognitive computing approach for rolling element bearing monitoring.
• Visual cognition principles enhance the robustness and accuracy of fault detection systems.
• This technique provides a valuable tool for predictive maintenance and industrial equipment reliability.