Fatigue damage diagnosis and prognosis for 2024 aluminum plates with center holes: A strain monitoring approach

A fatigue damage diagnosis and prognosis approach has been developed based on strain monitoring. The fatigue experiment is conducted on 2024 aluminum specimens with center holes to obtain the strain/crack information. The occurrence of cracks is detected by a thresholding method. A data-driven Gaussian process regression (GPR) algorithm is used to establish the relationship between the crack length and the strain characteristic parameters. A dynamic Bayesian network (DBN) is assembled by combining the GPR algorithm and Paris formula for real-time crack propagation predictions. An experiment has been carried out to illustrate the accuracy of the approach. The results indicate that the threshold method can detect fatigue cracks effectively. The GPR algorithm can accurately identify fatigue crack length and has higher recognition accuracy than other algorithms. The crack propagation prediction error and time consumption of DBN are much smaller than mechanics-based crack growth analysis.