An Artificial Neural Network-Based Algorithm for Evaluation of Fatigue Crack Propagation Considering Nonlinear Damage...

Area of Science:

• Aerospace Engineering
• Materials Science
• Computational Mechanics

Background:

• Damage tolerance is critical in aerospace and aviation.
• Fatigue crack growth modeling is complex due to nonlinear factors.
• Artificial Neural Networks (ANNs) show potential for nonlinear modeling.

Purpose of the Study:

• To propose a novel fatigue crack calculation algorithm using Radial Basis Function-ANN (RBF-ANN).
• To model the nonlinear relationship in fatigue crack growth using experimental data.
• To account for loading interaction effects using an equivalent stress intensity factor.

Main Methods:

• Developed a fatigue crack growth algorithm based on RBF-ANN.
• Utilized experimental data, including equivalent stress intensity factor, for training.
• Validated the model using constant amplitude loading, varying stress ratios, and overloads.
• Compared RBF-ANN results with Forman and Wheeler equations.

Main Results:

• The RBF-ANN model demonstrated superior agreement with experimental data compared to Forman and Wheeler equations.
• The proposed ANN-based approach effectively handles the complexities of fatigue crack growth.
• The model shows significant advantages in predicting crack propagation under various loading conditions.

Conclusions:

• RBF-ANN is a powerful and advantageous tool for modeling fatigue crack growth.
• The proposed algorithm offers a sophisticated and promising method for fatigue crack growth computation, especially considering loading interactions.
• This approach has significant implications for improving structural integrity and safety in aerospace.