An Intelligence Method for Recognizing Multiple Defects in Rail.

Area of Science:

• Materials Science
• Mechanical Engineering
• Non-destructive Testing

Background:

• Ultrasonic guided waves are effective for rail defect detection.
• Current algorithms often require complex feature extraction based on expert knowledge.
• A need exists for automated and robust rail defect identification methods.

Purpose of the Study:

• To propose and validate a new method for identifying multiple types of rail defects.
• To develop an automated feature extraction technique for guided wave signals.
• To enhance the reliability of rail integrity monitoring.

Main Methods:

• Utilized segment principal components analysis (S-PCA) for signal feature extraction.
• Employed a Support Vector Machine (SVM) model for defect classification.
• Conducted simulations and experiments on rails with various defects (crack, corrosion, transverse crack under shelling).
• Acquired guided wave signals using nine excitation-reception channels.

Main Results:

• The proposed method effectively identified different rail defects.
• High classification accuracy rates of 96.29% and 96.15% were achieved by combining multiple signals.
• The S-PCA and SVM approach demonstrated robustness in defect recognition.

Conclusions:

• The developed S-PCA and SVM method offers an effective solution for automated rail defect identification.
• Combining signals from multiple channels significantly improves classification accuracy.
• This technique enhances the safety and maintenance of rail infrastructure.