Fig. 1. Flow chart of fatigue crack prediction based on FBG
Fig. 2. Component size diagram
Fig. 3. Strain field nephograms at different stages of crack propagation. (a) Crack initiation; (b) initial crack propagation; (c) steady crack propagation; (d) crack approaching failure
Fig. 4. Fatigue crack growth monitoring experimental system based on FBG
Fig. 5. Overall schematic diagram of FBG demodulation system
Fig. 6. Fatigue tensile results of samples 1--4
Fig. 7. Schematic diagram of FBG sensor array
Fig. 8. Actual pasting position of FBG sensor array
Fig. 9. Fitting result of fatigue extension a-N curve
Fig. 10. Wavelength change measured by sensor FBG1
Fig. 11. Relationship between peak-to-peak wavelength and cyclic loading times
Fig. 12. Relationship between peak-to-peak wavelength and crack length
Fig. 13. Tensile results of aluminum alloy sample 5
Fig. 14. Principle diagram of GBRT
Fig. 15. Principle of 5-fold cross-verification
Fig. 16. Prediction result of GBRT algorithm
Parameter | Value |
---|
Density /(kg·m-3) | 2700 | Young’s modulus /MPa | 69 | Poisson’s ratio | 0.33 | Max principal stress /MPa | 84.4 | Coefficient of damage viscosity | 0.0001 |
|
Table 1. Material properties of simulation experiment
Number | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
---|
Cycle /104 | 0 | 3.0 | 3.5 | 4.0 | 4.5 | 5.0 | 5.5 | 6.0 | 6.1 | Crack length /mm | 0 | 2.2 | 2.6 | 3.1 | 4.4 | 6.2 | 7.8 | 13.3 | 21.0 |
|
Table 2. Results of fatigue tensile test
Evaluation index | SSE | MAE | MSE | R2 |
---|
Value | 0.0820 | 0.0713 | 0.0358 | 0.9912 |
|
Table 3. Evaluation indexes of fitting curve
Model | 0 | 1 | 2 | 3 | 4 |
---|
GBRT | -0.486222 | -0.279238 | -0.402655 | -0.632112 | -4.465716 |
|
Table 4. 5-fold cross-validation results of GBRT model
Model | EV | MAE | MSE | R2 |
---|
GBRT | 0.999947 | 0.018683 | 0.000699 | 0.999935 |
|
Table 5. Regression performance evaluation results of GBRT model
Model | EV | MAE | MSE | R2 |
---|
Linear | 0.999743 | 0.034250 | 0.002771 | 0.999743 | SVR | 0.999670 | 0.061197 | 0.004568 | 0.999577 | GBRT | 0.999947 | 0.018683 | 0.000699 | 0.999935 |
|
Table 6. Regression performance evaluation results of model