Research On Acoustic Emission Characteristics Of Tensile Damage Of Carbon Fiber Composites At High Temperature

Carbon fibre-reinforced polymer have been widely used in aerospace,chemical and other fields in recent years because of their excellent properties such as high specific strength,high specific stiffness,large specific modulus.With the continuous expansion of its application range,many composite parts are inevitably used in harsh working environments such as high temperature.Under the influence of temperature,its damage is bound to be more complex,affecting the usability and reliability.In this paper,numerical simulation,experimental research and deep learning framework are used to study the damage characteristics,damage prediction and fracture load prediction of composites under the influence of high temperature.Through the comparative analysis of the damage of CFRP laminates under different temperature,it is found that the degree of matrix damage and interlayer damage is closely related to the temperature.With the increase of temperature,the damage of matrix and interface gradually becomes heavier,from sporadic damage to large-area damage,and the temperature is inversely proportional to the fracture load of the specimen.The AE signal classification of CFRP Laminates at different temperatures is realized by k-means,and the damage pattern recognition is realized.It is determined that the damage at different temperatures includes matrix cracking,delamination,and fiber breakage.It is determined that the matrix cracking and interlayer damage become more after the temperature increases,resulting in the failure of the material in the case of less fiber breakage.On this basis,the further analysis of the signal is realized based on the deep learning framework,and the damage pattern recognition and prediction of CFRP laminates in high temperature environment is realized by using the fully connected neural network,with an accuracy of91.4%.The CWT image is used as the input of convolutional neural network to realize the damage prediction of high temperature environment with higher recognition accuracy.Finally,the neural network is used to predict the final fracture load of the specimen by collecting AE data under low load(50% fracture load),and the network is suitable for the prediction of tensile fracture load under different temperature,and the prediction error is less than 5%.The research results can provide support for the damage early warning of carbon fiber composite structures working in high temperature environment.