Fatigue Performance and Remaining Lifetime Prediction of Dented Thin Sheet Metals Using Data Driven Approach

During routine service, the wing skin and airframe of the aircraft are particularly prone to the damage resulting from the low-velocity impact of particles, such as hailstones, runway debris, wrench dropping, or collision of ground vehicles. Meanwhile, for aging aircraft whose service life has reached more than 75% of the designed service life , its structure has endured long-term periodic loads in flight, which may lead to the formation of fatigue micro-cracks, in this case even the presence of invisibile foreign object damage (FOD) will accelerate fatigue crack growth. However, current evaluation methods follow a conservative approach, which implies a lower accuracy in fatigue life prediction of components and higher cost related to maintenance of airframes throughout their service life. Therefore, the estimation of residual fatigue life become important, which could play a meaningful guiding role in the maintenance and repair of aging aircraft. In recent years, data driven approach, as an efficient tool to assess complex non-linear relationships, is gaining interest in the fatigue life prediction. Due to the mutual influence of each factor, it is difficult to analyze one factor independently without considering other influencing factors. Meantime, the increasing complexity of constitutive relationships in numerical material models make it more difficult to analyse the impact of the individual factors. The use of data driven approach could simplify the complexity of fatigue life prediction and provides a valuable solution for the fatigue life of dented thin plates. This project will research on the effect of dent followed by fatigue loading for thin plate. Advanced numerical modelling is applied to integrate the impact simulation in fatigue analysis and improve predictive accuracy using Digital Image Correlation (DIC). A data-set will be finally built to predict fatigue life of dented thin plate by using data driven approach.