Development of Maximum Residual Stress Prediction Technique for Shot-Peened Specimen Using Rayleigh Wave Dispersion...

Area of Science:

• Materials Science
• Mechanical Engineering
• Non-Destructive Testing

Background:

• Shot peening is a critical surface treatment for improving fatigue life and crack suppression in materials like nickel-based superalloys.
• Accurate evaluation of residual stress is essential for material integrity, especially in high-temperature, high-pressure environments (aerospace, nuclear power).
• Ultrasonic testing, specifically using Rayleigh waves, offers a non-destructive method for estimating residual stress.

Purpose of the Study:

• To predict the maximum compressive residual stress in shot-peened Inconel 718 specimens.
• To establish a relationship between Rayleigh wave dispersion and residual stress distribution.
• To develop and validate a convolutional neural network (CNN) model for this prediction.

Main Methods:

• Generated 173 residual stress distributions using Gaussian functions and factorial design.
• Converted stress distributions into 173 Rayleigh wave dispersion datasets for CNN training.
• Trained a convolutional neural network (CNN) model using the generated database and validated its performance.

Main Results:

• The CNN model successfully learned the relationship between Rayleigh wave dispersion data and residual stress distributions.
• The model demonstrated the ability to accurately predict the maximum compressive residual stress.
• Validation data confirmed the predictive performance of the developed CNN model.

Conclusions:

• The combined approach of Rayleigh wave dispersion analysis and CNN provides an effective non-destructive method for predicting residual stress.
• This technique is crucial for ensuring the structural integrity and safety of critical components made from shot-peened materials.
• The study highlights the potential of machine learning in non-destructive evaluation for materials science applications.