Agent Model Method For Material Parameter Recognition And Bubbling Development Of Test Equipment

For materials,their stability and reliability largely depend on the mechanical behavior of materials,and the mechanical behavior of materials is mainly affected b y the mechanical parameters of materials.Accurate mechanical parameters of materials are very important for the design and Simulation of functional devices,so the identification methods of mechanical parameters of materials have attracted muc h attention.However,for different kinds of materials,their constitutive models are diverse,while traditional experimental techniques or computational reverse engineering techniques can only be applied to specific models of specific materials,and the recognition and measurement accuracy have corresponding limitations.For example,the traditional measurement techniques such as indentation method and electrical measurement method have some limitations,such as fewer measuring points and greater environmental impact.Moreover,the Lenberg-Marquardt in the computational reverse engineering technology is restricted by the selection of initial values in the reverse engineering,so it is difficult to accurately obtain the constitutive model parameters of materials.Moreover,this kind of traditional identification method is difficult to identify the parameters of thin film materials and some non-linear materials.Therefore,a unified method can be studied as a way to identify the parameters of material constitutive model.This paper aims at the background and difficulties of the above-mentioned problems and carries out corresponding work.Its main work is as follows:(1)A new method of material constitutive model parameter identification based on surrogate model and bubbling experimental measurement technology is proposed.Firstly,the deflection experimental values of the corresponding measuring points of the material bubbling test specimens are measured by using the improved bubbling experimental device.Secondly,according to the corresponding bubbling experimental environment and the geometry and size of the specimens,the finite element model of the forward problem is established.Then,according to the procedure of agent model construction,the Latin hypercube experimental design method is selected to select a corresponding number of sample points of material parameters,and then the corresponding response values of material parameters are obtained by bringing the sample points into the positive problem model.The sample data set is constructed,and then the corresponding material polynomial agent model is constructed by combining the least square method.Finally,the inverse problem optimization model is constructed,and the sum of squares of the difference between the experimental deflection of the measured point and the proxy model of the corresponding measured point is taken as the objective function.The state transition algorithm is introduced to inverse the parameters,which realizes the double balance of the solving accuracy and the convergence speed of the inverse problem optimization.The agent model is used to transform the implicit non-linear objective function optimization problem into the explicit objective function optimization problem,which can avoid many problems in parameter identification.Because the agent model does not depend on experiments,the agent model database can be established in advance.It can lay an algorithm foundation for the instrumentation of bubbling test equipment,and can effectively solve the problem that it is difficult to unify the identification methods of mechanical parameters of thin film materials with different constitutive relations by bubbling method.(2)The proposed recognition method is validated accordingly.Firstly,304 stainless steel specimens with known material parameters and rubber materials with unknown material parameters were selected and made into bubble test specimens.The experimental deflection values of several measuring points were measured by improved bubble test device.Secondly,the constitutive model parameters of the two materials are inversely calculated by using the identification method in this paper,and the selection of the inverse interval of material parameters is emphatically discussed.In order to further confirm the accuracy of this method in identifying non-linear materials,two kinds of vulcanized rubber materials with different numbers of known material parameters are selected to verify.The experimental results show that this method can effectively and accurately identify non-linear materials with unknown mechanical parameters.(3)The basic scheme of developing bubbling test equipment is put forward,that is,the automatic parameter inversion identification s ystem is designed.In order to further solve the tedious operation of parameter identification process in this paper and achieve the purpose of batch processing of material parameters,based on the basic idea of identification in the previous paper,four modules in the whole identification process are sorted out,implemented by commercial software and programming language,and corresponding interfaces are provided,so that each module can not only run independently,but also run in batches efficiently and jointly.Processing improves work efficiency.Meanwhile,the sensitivity analysis of the agent model is carried out by using the global sensitivity analysis method.The influence of specimen thickness and bubbling test load on the agent model is effectively analyzed,which is conducive to the manufacture of standard specimens for bubbling test and the determination of loading system of the bubbling test device,and lays a foundation for the further establishment of the device test bench.