Application Of On-line Parameter Inversion Method For Large Structures

In the service process of large-scale structures,facilities and equipment,it is difficult to evaluate their condition due to the different service environments and complex stress states,such as large-scale steel structures,radar supports,and drilling platforms.At present,a large number of domestic and foreign scholars have studied a variety of detection methods and condition monitoring systems for this kind of problems.However,to accurately conduct overall condition assessment and simulation analysis of the research objects,constitutive parameters of structural dynamic changes are still needed.Accordingly,the combination of fast-developing artificial intelligence-related technologies has obvious advantages in data fitting,classification,and extraction,and processing large amounts of data.Therefore,this paper proposes a heuristic algorithm-based physical model numerical fitting material parameter inversion method.Under the premise of taking into account the calculation efficiency and accuracy,the material constitutive parameters in the complex and large-parameter model are identified,which is a large-scale structure.Condition monitoring,evaluation and prediction provide infrastructure parameter data,reduce real-time monitoring costs,and improve the accuracy of evaluation and prediction.The main work of this paper includes the following aspects:This paper first studies the fast parameter inversion method based on heuristic algorithm.Aiming at the problem of large-scale structure finite element simulation model,which has many nodes,large amount of parameters,complex response,and difficulty in algorithm implementation,a rapid parameter inversion method based on hill climbing algorithm is studied.After algorithm optimization and iteration,the measured true value is gradually approached.The constitutive parameters of the structure are obtained as a whole through a single operation,and the structure state is monitored,evaluated and predicted.The convergence speed and accuracy are taken into account by the global search first and then the classification evaluation,and the algorithm is verified by the hill climbing algorithm.Taking into account the mutual exclusion of algorithm training efficiency and accuracy,to further meet the needs of actual engineering applications,this paper studies the calculation of the finite element model of complex large-scale structural parts based on the proxy model of structural parts,which is used to find the balance point of efficiency and accuracy in engineering applications.Based on the artificial neural network fitting algorithm,this paper proposes a method to extract the representative basis vectors of the system response in the spatial dimension,and apply the radial basis function interpolation model to establish the nonlinear mapping relationship between arbitrary parameters and their corresponding ground state combination coefficients.The response field of the space-time degrees of freedom is modeled by proxy,matching the calculated value of the finite element model,and fitting the proxy model of the finite element model within a certain accuracy range,which greatly improves the calculation efficiency.Finally,the rapid parameter inversion based on the heuristic algorithm and the real-time structural state evaluation method studied in this paper are applied to the actual project of a large steel structure,and the effectiveness of this method is preliminarily verified.Provide research ideas and directions.