Research On Damage Of SiCp/Al Composites Based On Multi-Method Collaboration

Silicon carbide particle reinforced aluminum matrix composites(SiCp/Al)have attracted extensive attention in the fields of automobile manufacturing and electronic packaging because of their high specific strength,good thermal conductivity and dimensional stability.SiCp/Al composites with less than 20% SiC content can meet the strength requirements in addition to have high specific modulus and good wear resistance when they are used for structural materials.Thus,SiCp/Al composite is a lightweight structural metal material with great development potential.It is of great theoretical significance and engineering application value to study the failure form of SiCp/Al composites and analyze the damage mechanism during loading.In the past,researchers mostly used experiments and theoretical methods to study the damage characteristics of SiCp/Al materials.With the continuous development of computer science and the rapid rise of artificial intelligence,finite element simulation and neural network have been more and more widely used.The combination of experiments,numerical simulation and deep learning methods can effectively identify the failure form and damage mechanism of SiCp/Al materials and save a lot of time and cost.In this paper,the failure mode of SiCp/Al composites during mechanical loading is studied by collecting acoustic emission(AE)signals through experiments.The effects of particle strength,particle volume fraction and random distribution on the damage evolution of composites were studied by ABAQUS tensile finite element simulation,and the causes of material damage were analyzed.At the same time,the deep learning model is introduced to realize the damage area identification related to the microstructure characteristics of composites.The main work contents can be divided into three parts:1.Build an experimental platform to carry out uniaxial tensile test on SiCp/Al composite samples,collect AE signals during the tensile test through the monitoring platform,make time-frequency analysis,principal component analysis and fuzzy cluster analysis on the signals to characterize the damage mechanism of SiCp/Al composite,and identify three main damage modes of SiCp/Al composite: debonding of SiC/Al interface,Al fracture and SiC particle fracture.After the tensile test,the fracture surface was analyzed by scanning electron microscope(SEM),and the results of SEM images and energy spectrum analysis confirmed the acoustic emission results.2.The secondary development of ABAQUS is realized by Python language,and the finite element model considering the particle strength,particle volume fraction and random distribution of reinforcement is established to analyze the micromechanical mechanism of SiCp/Al composites.The results show that the influence of particle strength and random distribution on the stress-strain relationship before the material damage begins is not obvious,but it has great influence on the damage stage,maximum strength and corresponding failure strain of the material.With the increase of particle volume fraction,the damage intensity of the model becomes larger and larger;The macro response strength of composites with higher particle volume fraction is more sensitive;Particle aggregation and particle closing to the boundary will affect the damage evolution of the model,which is the cause of material damage.3.The deep learning model is introduced,and the data generated by finite element simulation is used to train and evaluate the network model to predict the stress distribution of SiCp/Al composites under uniaxial tensile load.The results show that,compared with the finite element results,the convolution neural network(CNN)method significantly improves the calculation efficiency.CNN model can identify the damage area related to the microstructure characteristics of composites,and make rapid prediction based on the given two-dimensional microstructure image.The network model can locate and visualize the important characteristics of stress distribution,especially the stress distribution on particles.