A Feature Extraction Method Using Improved Multi-Scale Entropy for Rolling Bearing Fault Diagnosis.

Area of Science:

• Mechanical Engineering
• Signal Processing
• Fault Diagnosis

Background:

• Rolling bearings are critical components in machinery.
• Effective fault diagnosis is essential for preventing catastrophic failures.
• Existing entropy-based methods face challenges like information leakage.

Purpose of the Study:

• To propose an improved multi-scale entropy (IMSE) method for rolling bearing fault diagnosis.
• To address information leakage issues in similarity calculations for machinery systems.
• To enhance the accuracy of fault feature extraction and identification.

Main Methods:

• Developed the improved multi-scale entropy (IMSE) feature extraction technique.
• Utilized Pythagorean Theorem and similarity criteria to overcome information leakage.
• Employed a support vector machine (SVM) classifier for feature identification.
• Conducted experiments on rolling bearings under various conditions.

Main Results:

• The IMSE method successfully extracts status information from rolling bearings.
• Experimental results demonstrate improved identification accuracy using IMSE features.
• IMSE outperforms traditional multi-scale entropy (MSE) and sample entropy (SE) methods.
• The proposed method effectively identifies bearing conditions.

Conclusions:

• The IMSE method is a robust approach for rolling bearing fault diagnosis.
• IMSE offers superior feature extraction and identification accuracy.
• This advancement contributes to more reliable machinery health monitoring.