Research On Technology And Application Of Combined Approximate Model Based On Radial Basis

Approximate model technology is a method of constructing models required for complex optimization problems.It can effectively solve the difficult problems of optimization model construction.It requires a small number of sample points,which can greatly reduce calculation costs and improve calculation efficiency;but in the face of increasingly complex problems Actual optimization problems,the prediction accuracy and calculation efficiency of approximate model technology are not enough to meet the needs of actual engineering.This paper studies a combined approximate model technology based on radial basis and applies it to actual optimization design problems to solve similar problems.Provide a new idea.The main work of this paper is as follows:(1)In order to build a high-precision approximate model,use test functions and accuracy evaluation indicators to test the performance of common approximate model technologies,and compare and analyze the fitting effects.The radial basis approximate model technology is selected as the research object,combined with experimental design Method and intelligent optimization algorithm,construct the research system framework of combined approximate model technology based on radial basis.(2)Aiming at the shortcomings of backtracking search algorithms that are easy to fall into local optimality and poor robustness,a backtracking search optimization algorithm based on combined mutation strategy is proposed.The performance simulation analysis of the improved algorithm is performed using test functions.The results show that the proposed algorithm is The improved algorithm has great advantages in convergence speed and accuracy;through the analysis and comparison of common experimental design methods,the Latin hypercube experimental design method is selected as the research object,and the improved backtracking search optimization algorithm is combined with the Latin hypercube design method.An improved Latin hypercube experimental design method based on the backtracking search optimization algorithm,and its performance simulation analysis shows that the improved Latin hypercube experimental design method improves the original method's defects such as uneven sampling and low stability,which is a follow-up approximation The acquisition of model technology sample points has laid a good foundation.(3)Aiming at the shortcomings of the radial basis approximation model technology in dealing with complex problems such as low accuracy and low efficiency,the radial basis approximation model technology is improved through the dynamic point addition strategy and the fusion basis function strategy,and a method based on multiple Based on the strategy of radial basis approximation model technology,its fitting performance is analyzed and compared.The results show that the proposed multi-strategy-based radial basis approximation model technology has higher accuracy and stability;at the same time,in order to obtain higher precision Approximate model,based on the improved radial basis approximation model technology,a combined approximate model technology based on radial basis is proposed,and its fitting performance is analyzed and compared.The results show that the combined approximate model technology based on radial basis has higher Prediction accuracy and solution efficiency.(4)The combined approximate model technology based on the radial basis is applied to the actual engineering optimization problem.The column of the SZJY-14 machining center is taken as the research object,and the structure is optimized.First,use the finite element analysis software Abaqus to mesh the three-dimensional structure of the column,and perform static and dynamic performance analysis to determine the optimization goal;second,use the sensitivity analysis method to determine the optimization design variables,and use the combined approximate model technology based on radial basis A column optimization model was established;finally,an improved backtracking search optimization algorithm was used to realize the column structure optimization design.The solution results show that compared with the original structure of the column,the optimized mass is reduced by 3.77%,the maximum deformation is reduced by 5.14%,and the first-order frequency is increased by 5.9%,achieving the expected effect.