Fatigue Crack Quantification Model for Metallic Structures Based on Strain Monitoring

The real-time monitoring of fatigue crack length is crucial for estimating the damage tolerance and the residual lifetime of aircraft metallic structures. This paper proposed a deep learning-based approach for predicting the fatigue crack size of metallic structures using strain monitoring data. By constructing learning models for Cycle Consistent Adversarial Networks, crack sizes are classified and quantified, and the correlation between the measured strain data and finite element simulation data was established. The proposed approach was applied for monitoring the growth of fatigue crack in a central hole plate subjected to random fatigue loads. The predictions show that the proposed model can monitor the fatigue crack lengths with high accuracy, where the prediction error of the crack length is less than 1 mm. This approach provides a reliable and accurate method for predicting the fatigue crack size of metallic structures, which may have important practical applications in aviation industry.