Research On Interlayer Strengthening And Damage Prediction Of CFRP Based On Acoustic Emission

Composite material is a kind of widely used new material.It has excellent properties such as high specific strength,high specific stiffness,corrosion resistance and low density,and has been applied in many fields,including automobile,aerospace,weapons and equipment.The damage of fiber reinforced composites is generally divided into two types: intralayer and interlayer.The intralayer damage mechanism refers to the damage in the same layer.The interlayer damage mainly refers to delamination.A large part of the failure of the composite material is caused by delamination.Due to the continuous expansion of the delamination,the strength of the material is finally reduced,resulting in serious consequences.Therefore,many structural designers pay more and more attention to the delamination of composite materials,and it is particularly important to predict the residual properties of materials as early as possible.Firstly,in this paper,carbon nanotubes(CNTs)layers were prepared on unidirectional and plain woven fabrics by electrostatic spraying.The content of CNTs was determined by controlling the spraying time,and CNTs film layers with different contents were obtained.CNTs can enhance the bonding strength between layers.The toughening mechanism and damage mode were analyzed by acoustic emission(AE)technique and optical microscope.The results show that the toughening mechanism of CNTs on different structures is obviously different.For unidirectional fabrics,the carbon nanotube layer effectively improves the fracture toughness of the interface,but for plain woven fabrics,in the case of low carbon nanotube area density,it only plays a role in the initial fracture toughness.The matrix cracking corresponding to the low frequency band of the AE signal is significantly increased.Combined with the observation of the fracture surface and the crack propagation path,the agglomeration of high content CNTs weakens the interlayer strength of the plain woven fabric,while the fracture toughness of the unidirectional fabric is significantly enhanced due to the increase of the crack propagation path,which also verifies the results of the acoustic emission.Secondly,the prediction model of residual interlayer performance is established by the AE signal obtained in the mode I experiment.Two feature selection algorithms,RRelief F and Neighbourhood component analysis,are used to evaluate the damage quantitative characterization ability of traditional AE features and cumulative AE features.It is found that cumulative features contain more information on the performance degradation of composite materials than non-cumulative features.Then,a Gated Recurrent Unit(GRU)network with temporal pattern attention(TPA)mechanism is established.The convolution operation is used to extract the internal correlation of multivariate time series,and compared with long short-term memory(LSTM)neural network,GRU and GRU based on traditional attention mechanism(GRU-Attention).Compared with the other three models,the prediction error of the TPA-GRU on the four data sets is the lowest,and the prediction performance of the TPA-GRU model is better than the other three traditional prediction models.