Fatigue Performance Prediction Study of Al-Li Alloy-based on Experimental and"Shallow"+"Deep"Hybrid Neural Network Methods
With its excellent mechanical properties,aluminum-lithium alloy has increasingly important in aerospace field and is one of the most rapidly developing lightweight materials.Fatigue and fracture are one of the main causes of failure of aerospace structural components,and fatigue damage is highly uncertain and sudden,so its fatigue performance evaluation and prediction has become a hot research topic.The fatigue experiments of 2A97 aluminum-lithium alloy considering the sampling direction,notch,etc.are carried out to obtain eight sets of complete S-N curves and the influence of relevant factors on the fatigue performance of aluminum-lithium alloy are analyzed.A hybrid neural network model based on shallow network and deep learning is innovatively proposed.The shallow algorithm is used to train the low cycle fatigue experimental data to realize data derivation at first,and then accurately predict the fatigue limit of aluminum-lithium alloy under different working conditions by deep learning.This method provides a new way for the research of fast and accurate evaluation of fatigue properties of materials.
万方数据
