Lamb Wave Damage Quantification Using GA-Based LS-SVM.

Area of Science:

• Materials Science
• Mechanical Engineering
• Non-Destructive Evaluation (NDE)

Background:

• Lamb waves are effective for NDE, but damage quantification remains challenging due to complex propagation and detection mechanisms.
• Existing data-driven methods require robust models for accurate crack size evaluation.

Purpose of the Study:

• To develop and validate a reliable Lamb wave-based damage quantification method for crack evaluation.
• To enhance the accuracy and robustness of non-destructive testing using advanced machine learning algorithms.

Main Methods:

• A least square support vector machine (LS-SVM) model was developed for damage quantification.
• A genetic algorithm (GA) was used to optimize the LS-SVM model parameters.
• Three damage-sensitive features (normalized amplitude, phase change, correlation coefficient) were extracted from Lamb wave signals.

Main Results:

• The GA-optimized LS-SVM model accurately evaluated crack sizes using the proposed damage-sensitive features.
• Validation with coupon and lap joint test data, including fatigue cracks, demonstrated the method's effectiveness.
• The approach showed robustness across different loading conditions and manufacturers.

Conclusions:

• The proposed GA-based LS-SVM method offers a robust and accurate solution for Lamb wave-based crack quantification in NDE.
• This technique improves the reliability of damage assessment in structural health monitoring applications.