Font Size: a A A

Radial Basis Function Metamodel Based On Combined Strategy And Its Application

Posted on:2019-02-28Degree:MasterType:Thesis
Country:ChinaCandidate:F Y LuFull Text:PDF
GTID:2322330566967534Subject:(degree of mechanical engineering)
Abstract/Summary:PDF Full Text Request
Metamodel is one of the effective ways to solve complicated engineering optimization problems,there is still lack of computational efficiency and prediction accuracy can not take into account but for complex high dimension problems.In order to improve the computational efficiency and accuracy of surrogate model technique,based on the theory of model technology and method,the Radial Basis Function(RBF)metamodel technology as the research object,construct the RBF metamodel technology research framework,put forward the RBF metamodel technology based on a combined strategy,and the optimization design of its application to practical engineering problems,aimed at providing a feasible solution for the similar practical problems.The main research work is as follows:(1)Aiming at the approximate optimization of complex engineering black-box problem,the performance analysis and comparison of several commonly used surrogate model techniques are carried out by using model evaluation criteria.The RBF metamodel is selected as the object,and the research framework of RBF metamodel is constructed by combining the experimental design and intelligent optimization algorithm.(2)The mechanism of several typical surrogate model methods:RBF method,Kriging method,and polynomial response surface method is studied.The numerical tests and model evaluation criteria are used to test and analyze the models and determine the RBF model as the research object of this paper;(3)Aiming at the deficiency of low prediction accuracy and efficiency of RBF model technology in dealing with complex problems,this paper improves the RBF by analyzing the sample point factors and structural factors that affect the prediction accuracy of the model.Considering the influence of sample point factors,an augmented RBF method based on multi-strategy(ARBF)is proposed.An example shows that the proposed method can ensure the calculation efficiency and prediction accuracy.By analyzing the structure of RBF,a new Hybrid RBF model based on kernel function(HRBF)is proposed.The experimental results show that the improved method can effectively overcome the structural risk of RBF and improve the prediction accuracy of the model.By utilizing the advantages of the two improved techniques,the RBF technology based on combined strategy(CORBF)is proposed to extend the fitting ability of the technology for complex high-dimensional problems.By numerical and engineering simulation analysis,compared with the former two improved techniques,this technique can not only guarantee the structural stability of RBF in the fitting process,but also have higher optimization efficiency and accuracy.The adaptability of RBF technology to high dimensional problems has also been extended.(4)Taking the vertical column of horizontal central milling machine as the research object,the research on the engineering application of CORBF technology is carried out.The model of column structure is established on the platform of Pro-E and Abaqus.The optimization objective is determined by statics and modal analysis,the design variable is determined by sensitivity analysis,and the optimization model of column is constructed.The CORBF technology is used in the optimization task.After optimization,the mass of the column is reduced by 4.8,the maximum deformation is reduced by 14.776 and the first order frequency is increased by 8.4744.The maximum deformation of the column is decreased by 14.776 and 8.474 respectively.It is shown that the proposed technology is of great theoretical significance and practical value to the optimization design of complex engineering problems.
Keywords/Search Tags:Radial basis function metamodel technology, Design of experiment, Seven-spot ladybird optimization algorithm, Combination strategy, Optimization design
PDF Full Text Request
Related items