Prediction Of Aerostructural Dynamic Response And Analysis Of Vibration Fatigue Life Based On LSTM

In the aviation industry,structure subject to complex vibration loads are prone to fatigue damage.Before fatigue damage,there is no obvious deformation in the structure.Once fatigue damage occurs,it cannot be reversed,which is extremely harmful to the structure.Therefore,fast calculation of structural fatigue damage and accurate prediction of structural fatigue life have important theoretical significance and engineering value for ensuring the safe operation of aeronautical structures.Among the vibration fatigue life analysis methods,compared with the frequency domain method,the time domain method has higher accuracy and wider application range.It is more accord with the requirements of fatigue analysis of aerostructures under high frequency vibration loads.However,the time domain method needs to obtain a complete stress time history.In order to solve the problem of calculation quantity,it is very important to propose a method for quickly predicting the structural dynamic response.In recent years,deep neural networks have attracted widespread attention due to their powerful nonlinear fitting capabilities,especially for their good feature extraction and prediction capabilities for time series data,which provided new idea for vibration fatigue life analysis of aeronautical structures.The main content of this paper is as follows:(1)The vibration fatigue life analysis methods are introduced,including the frequency domain method and the time domain method.(2)A method for predicting structural dynamic response under random vibration is proposed.In order to accurately predict the high-frequency vibration fatigue life of aeronautical structures,taking advantage of the ability of Long Short-Term Memory(LSTM)neural network which can effectively extract the relationship between random vibration and dynamic response.It can realize the fast calculation of structural dynamic response and solve the calculation problem of time-domain method.(3)Cantilever beam is a common component in aeronautical structures,the prediction method in this paper is used to conduct a comprehensive analysis of the defective cantilever beam model.Random vibration loads are added to the finite element model of the defective cantilever beam,the dynamic response of the dangerous position is extracted,which is divided into training samples and verification samples.The LSTM neural network is trained and verified,and the trained neural network can accurately predict the long-term structural dynamic response based on random vibration.Based on the results of the LSTM neural network,the fatigue life of the structure is calculated and compared with the expreimental result.The result shows that the method in this paper can quickly and accurately calculate the vibration fatigue life of the structure.(4)Based on the dynamic response prediction method in this paper,the vibration response prediction and fatigue life calculation of the aircraft pod are discussed.The aircraft pod is a common load-bearing structure,and there is a phenomenon of stress concentration in the bolt holes inside the cabin,which is more prone to fatigue damage under the vibration load.The LSTM neural network is used to predict the dynamic response and fatigue life of the aircraft pod.The result shows that the prediction method in this paper can improve the computational efficiency of the time-domain method,which can also provide assistance for the high-frequency vibration fatigue life analysis of aeronautical structure and lay the foundation for the subsequent analysis of structural nonlinear vibration fatigue.