Improvement And Application Of Reduce Basis Method For Parameter Separation

At present,the engineering activities are increasingly fierce,and the rapid development of modern society has a higher standard for shortening the design cycle in structural analysis.In practical engineering problems,the owner often requires the designer to design the engineering structure quickly.With the increasing complexity of large-scale structures,the use of finite element analysis to determine the optimal solution of the structure affects the efficiency of optimal design.Therefore,it is necessary to use the limited computer resources to calculate and solve the structure effectively and quickly.Over the years,experts and scholars at home and abroad have studied a lot of basic theories of algorithms,and put forward many algorithms with sufficient accuracy and computational efficiency.However,there are still many problems to be solved.In this paper,with ANSYS software,we extract the required linear elastic operator,and use the inverse reduced basis method to get the linear elastic operator which is independent of the design parameters.The process of parameter separation in the reduced basis method is improved.Based on the improved reduced basis method,the fast calculation of frame structure and underground side wall structure is realized,and its parameters are optimized by combining with genetic algorithm.In this paper,theoretical analysis and numerical simulation are combined.Firstly,the parameter separation of element stiffness matrix is studied theoretically.On this basis,a simple model of the same element composition is established.The relevant matrix information is extracted by ANSYS,and the linear elastic operator independent of design parameters is obtained by inverse reduced basis method.Secondly,the improved reduced basis method is applied to the plane frame structure and the underground side wall structure.After the improved reduced basis method,In the off-line stage,the uniform distribution is used to select the sample parameters,and then the static sample space is constructed by the corresponding finite element displacement of these parameters.In the on-line stage,the linear elastic matrix independent of the design parameters is calculated,and the new design parameters are selected,which can be used for real-time calculation.Thirdly,taking the finite element solution as the real solution,the relative error based on the improved reduced basis method is obtained,which has higher calculation accuracy.Different numbers of displacement solutions are selected to form reduced basis space.The accuracy of the improved reduced basis space is tested by the asymptotic method.The output of the results is calculated.The calculation error efficiency of the structure is solved.The variation of the relative error and error efficiency is analyzed.Then,the time consumed in the calculation process is compared between the finite element software and the improved reduced basis method.The improved reduced basis method can carry out real-time calculation at the present stage and significantly shorten the calculation process time.In order to ensure the accuracy and efficiency of calculation,the number of parameter samples in solution space is selected reasonably.Finally,the online phase of the improved reduced basis method is combined with genetic algorithm to optimize the parameters of frame structure and underground side wall structure.In the iterative process,there is no need to carry out the finite element analysis again according to the change of parameters.The improved reduced basis method is used for the fast calculation of the structure,which improves the optimization speed in the structural design and reduces the tedious time.The improved reduced basis method has higher accuracy,faster speed for structural solution and optimization analysis,and higher calculation efficiency.