Reverse Identification Method Of Plate Damage And Strength Based On Ball Head Expansion Data

Plates may have different mechanical properties due to different production technologies,production batches,and rolling thicknesses,and the mechanical properties of the structural materials in service will fluctuate with time or temperature.In addition,traditional uniaxial tension not only requires applying a load in the clamping area,which reduces the material utilization rate,but also tends to generate wrinkled areas during deformation of large specimens of thin plate materials,which can interfere with testing.Drawing on traditional testing methods such as tensile testing,providing a fast and resource saving method for testing the mechanical properties of materials,which is an urgent technology to solve the testing of materials such as aluminum plastic for new energy batteries,thin stainless steel,and nuclear power transducers.Therefore,it is important to reverse identification material parameters such as damage and strength based on the bulging deformation of small specimens in this paper.In this paper,the hardening index,damage parameters and tensile strength of materials are studied by means of bulging test and finite element simulation test,and the prediction and identification methods of correlation method and deep learning method are proposed.The correlation method is to customize the simulation material,fit the correlation equation,and reverse predict the real performance parameters of the material;the deep learning method is to identify the data and picture features of the input layer through the convolution neural network,adjust the parameters of the hidden layer and the output layer,make the predicted value of the parameters close to the real value,and reverse identify the performance parameters of the material.In this paper,the bulging test is carried out with the replaceable bulging test device of the male and female dies,and the influence of the die size on the bulging load-displacement curve is analyzed;based on the principle of equivalent energy,the hardening parameters of the material are obtained by correlation method,and the stress-strain curve of the material is obtained reversely;based on the GTN damage model,the influence of damage parameters on each stage of bulging is studied.Using the deep learning algorithm,the load-displacement curve obtained from bulging simulation test is taken as the input layer of the convolution nerve network in the form of pictures,and the damage parameters of the material are inversely identified;the tensile strength of the material is inversely obtained by correlation method and depth learning method,and the relationship between bulging load and tensile strength is constructed by correlation method.The tensile strength of the material is inversely recognized by using deep learning method to extract data features through one-dimensional convolution and two-dimensional convolution to extract image features.